{ "issues": [ { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/186", "id": 3802056590, "node_id": "I_kwDOPdnatc7inseO", "number": 186, "title": "Suggestion", "user": { "login": "JonathanAlkalay", "id": 87444377, "node_id": "MDQ6VXNlcjg3NDQ0Mzc3", "avatar_url": "https://avatars.githubusercontent.com/u/87444377?v=4", "gravatar_id": "", "url": "https://api.github.com/users/JonathanAlkalay", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2026-01-11T21:13:16Z", "updated_at": "2026-01-11T21:13:16Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Add sage attention as it speeds up inference by 2 to 3 times without quality loss", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/186/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/185", "id": 3794576380, "node_id": "I_kwDOPdnatc7iLKP8", "number": 185, "title": "running InifiteTalk-I2V-GGUF_lip-sync", "user": { "login": "JARRADSCHUETZ", "id": 68444231, "node_id": "MDQ6VXNlcjY4NDQ0MjMx", "avatar_url": "https://avatars.githubusercontent.com/u/68444231?v=4", "gravatar_id": "", "url": "https://api.github.com/users/JARRADSCHUETZ", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2026-01-08T21:52:41Z", "updated_at": "2026-01-08T22:39:14Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "\"Image\"\n\n^^ when running the workflow, it gets to the \"WANVIDEO SAMPLER\" node @ 86% then the error appears. When I look into the folder, it appears the JSON file is there but, a new temp file is being created every single time resulting in the error. How can I go about fixing this?\n\n\"Image\"\n\nlog\nAdding extra search path custom_nodes C:\\Users\\JARRA\\Documents\\ComfyUI\\custom_nodes\nAdding extra search path download_model_base C:\\Users\\JARRA\\Documents\\ComfyUI\\models\nAdding extra search path custom_nodes C:\\Users\\JARRA\\OneDrive\\Documents\\ComfyUI\\resources\\ComfyUI\\custom_nodes\nSetting output directory to: C:\\Users\\JARRA\\Documents\\ComfyUI\\output\nSetting input directory to: C:\\Users\\JARRA\\Documents\\ComfyUI\\input\nSetting user directory to: C:\\Users\\JARRA\\Documents\\ComfyUI\\user\n[START] Security scan\n[DONE] Security scan\n** ComfyUI startup time: 2026-01-08 16:33:18.744\n** Platform: Windows\n** Python version: 3.12.11 (main, Aug 18 2025, 19:17:54) [MSC v.1944 64 bit (AMD64)]\n** Python executable: C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Scripts\\python.exe\n** ComfyUI Path: C:\\Users\\JARRA\\OneDrive\\Documents\\ComfyUI\\resources\\ComfyUI\n** ComfyUI Base Folder Path: C:\\Users\\JARRA\\OneDrive\\Documents\\ComfyUI\\resources\\ComfyUI\n** User directory: C:\\Users\\JARRA\\Documents\\ComfyUI\\user\n** ComfyUI-Manager config path: C:\\Users\\JARRA\\Documents\\ComfyUI\\user\\__manager\\config.ini\n** Log path: C:\\Users\\JARRA\\Documents\\ComfyUI\\user\\comfyui.log\n[ComfyUI-Manager] Skipped fixing the 'comfyui-frontend-package' dependency because the ComfyUI is outdated.\n[PRE] ComfyUI-Manager\nCheckpoint files will always be loaded safely.\nC:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\cuda\\__init__.py:283: UserWarning: \n Found GPU1 NVIDIA GeForce GTX 1080 which is of cuda capability 6.1.\n Minimum and Maximum cuda capability supported by this version of PyTorch is\n (7.5) - (12.0)\n \n warnings.warn(\nC:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\cuda\\__init__.py:304: UserWarning: \n Please install PyTorch with a following CUDA\n configurations: 12.6 following instructions at\n https://pytorch.org/get-started/locally/\n \n warnings.warn(matched_cuda_warn.format(matched_arches))\nC:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\cuda\\__init__.py:326: UserWarning: \nNVIDIA GeForce GTX 1080 with CUDA capability sm_61 is not compatible with the current PyTorch installation.\nThe current PyTorch install supports CUDA capabilities sm_75 sm_80 sm_86 sm_90 sm_100 sm_120.\nIf you want to use the NVIDIA GeForce GTX 1080 GPU with PyTorch, please check the instructions at https://pytorch.org/get-started/locally/\n\n warnings.warn(\nTotal VRAM 12288 MB, total RAM 32624 MB\npytorch version: 2.9.1+cu130\nSet vram state to: NORMAL_VRAM\nDevice: cuda:0 NVIDIA GeForce RTX 3080 Ti : cudaMallocAsync\nUsing async weight offloading with 2 streams\nEnabled pinned memory 14680.0\nworking around nvidia conv3d memory bug.\nFound comfy_kitchen backend eager: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'scaled_mm_nvfp4']}\nFound comfy_kitchen backend triton: {'available': True, 'disabled': True, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8']}\nFound comfy_kitchen backend cuda: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'scaled_mm_nvfp4']}\nUsing pytorch attention\nPython version: 3.12.11 (main, Aug 18 2025, 19:17:54) [MSC v.1944 64 bit (AMD64)]\nComfyUI version: 0.8.2\n[Prompt Server] web root: C:\\Users\\JARRA\\OneDrive\\Documents\\ComfyUI\\resources\\ComfyUI\\web_custom_versions\\desktop_app\n[START] ComfyUI-Manager\n[ComfyUI-Manager] network_mode: public\n[ComfyUI-Manager] The matrix sharing feature has been disabled because the `matrix-nio` dependency is not installed.\n To use this feature, please run the following command:\n C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Scripts\\python.exe -m pip install matrix-nio\n\nTotal VRAM 12288 MB, total RAM 32624 MB\npytorch version: 2.9.1+cu130\nSet vram state to: NORMAL_VRAM\nDevice: cuda:0 NVIDIA GeForce RTX 3080 Ti : cudaMallocAsync\nUsing async weight offloading with 2 streams\nEnabled pinned memory 14680.0\n'resemble-perth' not found. Watermarking will be unavailable.\n[MultiGPU Core Patching] Patching mm.soft_empty_cache for Comprehensive Memory Management (VRAM + CPU + Store Pruning)\n[MultiGPU Core Patching] Patching mm.get_torch_device, mm.text_encoder_device, mm.unet_offload_device\n[MultiGPU DEBUG] Initial current_device: cuda:0\n[MultiGPU DEBUG] Initial current_text_encoder_device: cuda:0\n[MultiGPU DEBUG] Initial current_unet_offload_device: cpu\n[MultiGPU] Initiating custom_node Registration. . .\n-----------------------------------------------\ncustom_node Found Nodes\n-----------------------------------------------\nComfyUI-LTXVideo N 0\nComfyUI-Florence2 N 0\nComfyUI_bitsandbytes_NF4 N 0\nx-flux-comfyui N 0\nComfyUI-MMAudio N 0\nComfyUI-GGUF N 0\nPuLID_ComfyUI N 0\nComfyUI-WanVideoWrapper Y 20\n-----------------------------------------------\n[MultiGPU] Registration complete. Final mappings: CheckpointLoaderAdvancedMultiGPU, CheckpointLoaderAdvancedDisTorch2MultiGPU, UNetLoaderLP, UNETLoaderMultiGPU, VAELoaderMultiGPU, CLIPLoaderMultiGPU, DualCLIPLoaderMultiGPU, TripleCLIPLoaderMultiGPU, QuadrupleCLIPLoaderMultiGPU, CLIPVisionLoaderMultiGPU, CheckpointLoaderSimpleMultiGPU, ControlNetLoaderMultiGPU, DiffusersLoaderMultiGPU, DiffControlNetLoaderMultiGPU, UNETLoaderDisTorch2MultiGPU, VAELoaderDisTorch2MultiGPU, CLIPLoaderDisTorch2MultiGPU, DualCLIPLoaderDisTorch2MultiGPU, TripleCLIPLoaderDisTorch2MultiGPU, QuadrupleCLIPLoaderDisTorch2MultiGPU, CLIPVisionLoaderDisTorch2MultiGPU, CheckpointLoaderSimpleDisTorch2MultiGPU, ControlNetLoaderDisTorch2MultiGPU, DiffusersLoaderDisTorch2MultiGPU, DiffControlNetLoaderDisTorch2MultiGPU, LoadWanVideoT5TextEncoderMultiGPU, WanVideoTextEncodeMultiGPU, WanVideoTextEncodeCachedMultiGPU, WanVideoTextEncodeSingleMultiGPU, WanVideoVAELoaderMultiGPU, WanVideoTinyVAELoaderMultiGPU, WanVideoBlockSwapMultiGPU, WanVideoImageToVideoEncodeMultiGPU, WanVideoDecodeMultiGPU, WanVideoModelLoaderMultiGPU, WanVideoSamplerMultiGPU, WanVideoVACEEncodeMultiGPU, WanVideoEncodeMultiGPU, LoadWanVideoClipTextEncoderMultiGPU, WanVideoClipVisionEncodeMultiGPU, WanVideoControlnetLoaderMultiGPU, FantasyTalkingModelLoaderMultiGPU, Wav2VecModelLoaderMultiGPU, WanVideoUni3C_ControlnetLoaderMultiGPU, DownloadAndLoadWav2VecModelMultiGPU\nWanVideoWrapper WARNING: FantasyPortrait nodes not available: No module named 'onnxruntime'\n\nImport times for custom nodes:\n 0.0 seconds: C:\\Users\\JARRA\\OneDrive\\Documents\\ComfyUI\\resources\\ComfyUI\\custom_nodes\\websocket_image_save.py\n 0.0 seconds: C:\\Users\\JARRA\\Documents\\ComfyUI\\custom_nodes\\ComfyUI-MelBandRoFormer\n 0.0 seconds: C:\\Users\\JARRA\\Documents\\ComfyUI\\custom_nodes\\comfyui-multigpu\n 0.1 seconds: C:\\Users\\JARRA\\Documents\\ComfyUI\\custom_nodes\\comfyui-kjnodes\n 0.1 seconds: C:\\Users\\JARRA\\Documents\\ComfyUI\\custom_nodes\\comfyui-videohelpersuite\n 0.7 seconds: C:\\Users\\JARRA\\Documents\\ComfyUI\\custom_nodes\\ComfyUI-WanVideoWrapper\n 2.0 seconds: C:\\Users\\JARRA\\Documents\\ComfyUI\\custom_nodes\\ComfyUI-ChatterboxTTS\n\nFailed to initialize database. Please ensure you have installed the latest requirements. If the error persists, please report this as in future the database will be required: (sqlite3.OperationalError) unable to open database file\n(Background on this error at: https://sqlalche.me/e/20/e3q8)\nStarting server\n\nTo see the GUI go to: http://127.0.0.1:8000\ncomfyui-frontend-package not found in requirements.txt\n[DEPRECATION WARNING] Detected import of deprecated legacy API: /extensions/core/widgetInputs.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.\ncomfyui-frontend-package not found in requirements.txt\ngot prompt\nC:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\functional.py:681: UserWarning: A window was not provided. A rectangular window will be applied,which is known to cause spectral leakage. Other windows such as torch.hann_window or torch.hamming_window are recommended to reduce spectral leakage.To suppress this warning and use a rectangular window, explicitly set `window=torch.ones(n_fft, device=)`. (Triggered internally at C:\\actions-runner\\_work\\pytorch\\pytorch\\pytorch\\aten\\src\\ATen\\native\\SpectralOps.cpp:842.)\n return _VF.stft( # type: ignore[attr-defined]\nConverted mono input to stereo.\nResampling input 24000 to 44100\nProcessing chunks: 100%|██████████| 2/2 [00:00<00:00, 2.38it/s]\n[MultiTalk] --- Raw speaker lengths (samples) ---\n speaker 1: 124160 samples (shape: torch.Size([1, 1, 124160]))\n[MultiTalk] Audio duration (194 frames) is shorter than requested (400 frames). Using 194 frames.\n[MultiTalk] total raw duration = 7.760s\n[MultiTalk] multi_audio_type=para | final waveform shape=torch.Size([1, 1, 124160]) | length=124160 samples | seconds=7.760s (expected max of raw)\nT5Encoder: 100%|██████████| 24/24 [00:00<00:00, 68.24it/s]\nprompt token count: tensor([12], device='cuda:0')\nT5Encoder: 100%|██████████| 24/24 [00:00<00:00, 294.20it/s]\nprompt token count: tensor([98], device='cuda:0')\n[MultiGPU Core Patching] text_encoder_device_patched returning device: cuda:0 (current_text_encoder_device=cuda:0)\nRequested to load CLIPVisionModelProjection\nloaded completely; 8339.57 MB usable, 1208.10 MB loaded, full load: True\nClip embeds shape: torch.Size([1, 257, 1280]), dtype: torch.float32\nCombined clip embeds shape: torch.Size([1, 257, 1280])\nCUDA Compute Capability: 8.6\nDetected model in_channels: 36\nModel cross attention type: i2v, num_heads: 40, num_layers: 40\nModel variant detected: i2v_480\nInfiniteTalk detected, patching model...\nmodel_type FLOW\nLoading LoRA: lightx2v_I2V_14B_480p_cfg_step_distill_rank64_bf16 with strength: 1.0\nUsing GGUF to load and assign model weights to device...\nLoading transformer parameters to cuda:0: 100%|██████████| 1633/1633 [00:00<00:00, 11428.58it/s]\n------- Scheduler info -------\nTotal timesteps: tensor([999, 961, 886, 800, 725, 687], device='cuda:0')\nUsing timesteps: tensor([999, 961, 886, 800, 725, 687], device='cuda:0')\nUsing sigmas: tensor([1.0000, 0.9616, 0.8866, 0.8008, 0.7259, 0.6875, 0.0000])\n------------------------------\nsigmas: tensor([1.0000, 0.9616, 0.8866, 0.8008, 0.7259, 0.6875, 0.0000])\nMultitalk audio features shapes (per speaker): [(194, 12, 768)]\nRope function: comfy\nMultitalk mode: infinitetalk\nSampling 194 frames in 3 windows, at 480x832 with 6 steps\nFETCH ComfyRegistry Data [DONE]\n[ComfyUI-Manager] default cache updated: https://api.comfy.org/nodes\nFETCH DATA from: C:\\Users\\JARRA\\Documents\\ComfyUI\\user\\__manager\\cache\\1514988643_custom-node-list.json [DONE]\n[ComfyUI-Manager] All startup tasks have been completed.\nWanVAE encoded input:torch.Size([1, 3, 81, 832, 480]) to torch.Size([1, 32, 21, 104, 60])\n[WanVAE encode] Allocated memory: memory=7.142 GB\n[WanVAE encode] Max allocated memory: max_memory=10.316 GB\n[WanVAE encode] Max reserved memory: max_reserved=12.312 GB\nWanVAE encoded input:torch.Size([1, 3, 1, 832, 480]) to torch.Size([1, 32, 1, 104, 60])\n[WanVAE encode] Allocated memory: memory=6.948 GB\n[WanVAE encode] Max allocated memory: max_memory=7.952 GB\n[WanVAE encode] Max reserved memory: max_reserved=12.125 GB\nSampling audio indices 0-81: 0%| | 0/6 [00:00= tmp2\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp4 = tl.full([1], 4, tl.int64)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp5 = tmp1 < tmp4\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp6 = tl.load(in_ptr1 + (4 + 144*x2 + (x1)), tmp5, eviction_policy='evict_last', other=0.0)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp7 = tl.full([1], 63, tl.uint8)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp8 = tmp6 & tmp7\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp9 = tl.full(tmp8.shape, 0.0, tmp8.dtype)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp10 = tl.where(tmp5, tmp8, tmp9)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp11 = tmp1 >= tmp4\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp12 = tl.full([1], 8, tl.int64)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp13 = tmp1 < tmp12\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp14 = tl.load(in_ptr1 + (12 + 144*x2 + ((-4) + x1)), tmp11, eviction_policy='evict_last', other=0.0)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp15 = tl.full([1], 15, tl.uint8)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp16 = tmp14 & tmp15\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp17 = tl.load(in_ptr1 + (4 + 144*x2 + ((-4) + x1)), tmp11, eviction_policy='evict_last', other=0.0)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp18 = tl.full([1], 2, tl.uint8)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp19 = tmp17 >> tmp18\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp20 = tl.full([1], 48, tl.uint8)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp21 = tmp19 & tmp20\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp22 = tmp16 | tmp21\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp23 = tl.full(tmp22.shape, 0.0, tmp22.dtype)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp24 = tl.where(tmp11, tmp22, tmp23)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp25 = tl.where(tmp5, tmp10, tmp24)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp26 = tmp25.to(tl.float32)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp27 = tmp0 * tmp26\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp29 = ((((x6 // 32) % 8)) % 2)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp30 = tl.full([1], 1, tl.int64)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp31 = tmp29 < tmp30\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp32 = tl.full([1], 0, tl.uint8)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp33 = tl.full([1], 4, tl.uint8)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp34 = tl.where(tmp31, tmp32, tmp33)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp35 = tmp28 >> tmp34\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp36 = tl.full([1], 15, tl.uint8)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp37 = tmp35 & tmp36\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp38 = tmp37.to(tl.float32)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp39 = tmp27 * tmp38\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp41 = ((x3 % 256)) // 32\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp42 = tmp41 >= tmp2\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp43 = tmp41 < tmp4\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp44 = tl.load(in_ptr1 + (8 + 144*(x3 // 256) + 2880*x4 + (((x3 % 256)) // 32)), tmp43, eviction_policy='evict_last', other=0.0)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp45 = tl.full([1], 63, tl.uint8)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp46 = tmp44 & tmp45\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp47 = tl.full(tmp46.shape, 0.0, tmp46.dtype)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp48 = tl.where(tmp43, tmp46, tmp47)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp49 = tmp41 >= tmp4\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp50 = tmp41 < tmp12\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp51 = tl.load(in_ptr1 + (12 + 144*(x3 // 256) + 2880*x4 + ((-4) + (((x3 % 256)) // 32))), tmp49, eviction_policy='evict_last', other=0.0)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp52 = tl.full([1], 4, tl.uint8)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp53 = tmp51 >> tmp52\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp54 = tl.load(in_ptr1 + (8 + 144*(x3 // 256) + 2880*x4 + ((-4) + (((x3 % 256)) // 32))), tmp49, eviction_policy='evict_last', other=0.0)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp55 = tl.full([1], 2, tl.uint8)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp56 = tmp54 >> tmp55\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp57 = tl.full([1], 48, tl.uint8)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp58 = tmp56 & tmp57\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp59 = tmp53 | tmp58\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp60 = tl.full(tmp59.shape, 0.0, tmp59.dtype)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp61 = tl.where(tmp49, tmp59, tmp60)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp62 = tl.where(tmp43, tmp48, tmp61)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp63 = tmp62.to(tl.float32)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp64 = tmp40 * tmp63\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tmp65 = tmp39 - tmp64\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] tl.store(in_out_ptr0 + (x6), tmp65, None)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] \nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] metadata: {'signature': {'in_out_ptr0': '*fp16', 'in_ptr0': '*fp16', 'in_ptr1': '*u8', 'in_ptr2': '*fp16', 'xnumel': 'i32', 'XBLOCK': 'constexpr'}, 'device': 0, 'constants': {'XBLOCK': 1024}, 'configs': [{(0,): [['tt.divisibility', 16]], (1,): [['tt.divisibility', 16]], (2,): [['tt.divisibility', 16]], (4,): [['tt.divisibility', 16]]}], 'device_type': 'cuda', 'num_warps': 4, 'num_stages': 1, 'debug': True, 'cc': 86}\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] Traceback (most recent call last):\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py\", line 778, in _precompile_config\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] binary = triton.compile(*compile_args, **compile_kwargs)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\triton\\compiler\\compiler.py\", line 349, in compile\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] fn_cache_manager.put_group(metadata_filename, metadata_group)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\triton\\runtime\\cache.py\", line 96, in put_group\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] return self.put(grp_contents, grp_filename, binary=False)\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\triton\\runtime\\cache.py\", line 117, in put\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] with open(temp_path, mode) as f:\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] ^^^^^^^^^^^^^^^^^^^^^\nE0108 16:35:26.795000 24568 Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py:780] [0/0_1] FileNotFoundError: [Errno 2] No such file or directory: 'C:\\\\Users\\\\JARRA\\\\AppData\\\\Local\\\\Temp\\\\torchinductor_JARRA\\\\triton\\\\0\\\\YGMCYRRSOS4PIDJL6VS3N4XFTCS4MXREF32NRXHKMJAU7DJDQQNQ\\\\tmp.26fc6c94\\\\__grp__triton_poi_fused___rshift___apply_lora_bitwise_and_bitwise_or_cat_lift_fresh_mul_select_split_split_with_sizes_sub_view_2.json'\nError during model prediction: FileNotFoundError: [Errno 2] No such file or directory: 'C:\\\\Users\\\\JARRA\\\\AppData\\\\Local\\\\Temp\\\\torchinductor_JARRA\\\\triton\\\\0\\\\YGMCYRRSOS4PIDJL6VS3N4XFTCS4MXREF32NRXHKMJAU7DJDQQNQ\\\\tmp.26fc6c94\\\\__grp__triton_poi_fused___rshift___apply_lora_bitwise_and_bitwise_or_cat_lift_fresh_mul_select_split_split_with_sizes_sub_view_2.json'\n\nSet TORCHDYNAMO_VERBOSE=1 for the internal stack trace (please do this especially if you're reporting a bug to PyTorch). For even more developer context, set TORCH_LOGS=\"+dynamo\"\n\nError during sampling: FileNotFoundError: [Errno 2] No such file or directory: 'C:\\\\Users\\\\JARRA\\\\AppData\\\\Local\\\\Temp\\\\torchinductor_JARRA\\\\triton\\\\0\\\\YGMCYRRSOS4PIDJL6VS3N4XFTCS4MXREF32NRXHKMJAU7DJDQQNQ\\\\tmp.26fc6c94\\\\__grp__triton_poi_fused___rshift___apply_lora_bitwise_and_bitwise_or_cat_lift_fresh_mul_select_split_split_with_sizes_sub_view_2.json'\n\nSet TORCHDYNAMO_VERBOSE=1 for the internal stack trace (please do this especially if you're reporting a bug to PyTorch). For even more developer context, set TORCH_LOGS=\"+dynamo\"\n\n!!! Exception during processing !!! FileNotFoundError: [Errno 2] No such file or directory: 'C:\\\\Users\\\\JARRA\\\\AppData\\\\Local\\\\Temp\\\\torchinductor_JARRA\\\\triton\\\\0\\\\YGMCYRRSOS4PIDJL6VS3N4XFTCS4MXREF32NRXHKMJAU7DJDQQNQ\\\\tmp.26fc6c94\\\\__grp__triton_poi_fused___rshift___apply_lora_bitwise_and_bitwise_or_cat_lift_fresh_mul_select_split_split_with_sizes_sub_view_2.json'\n\nSet TORCHDYNAMO_VERBOSE=1 for the internal stack trace (please do this especially if you're reporting a bug to PyTorch). For even more developer context, set TORCH_LOGS=\"+dynamo\"\n\nTraceback (most recent call last):\n File \"C:\\Users\\JARRA\\OneDrive\\Documents\\ComfyUI\\resources\\ComfyUI\\execution.py\", line 518, in execute\n output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\OneDrive\\Documents\\ComfyUI\\resources\\ComfyUI\\execution.py\", line 329, in get_output_data\n return_values = await _async_map_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\OneDrive\\Documents\\ComfyUI\\resources\\ComfyUI\\execution.py\", line 303, in _async_map_node_over_list\n await process_inputs(input_dict, i)\n File \"C:\\Users\\JARRA\\OneDrive\\Documents\\ComfyUI\\resources\\ComfyUI\\execution.py\", line 291, in process_inputs\n result = f(**inputs)\n ^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\custom_nodes\\ComfyUI-WanVideoWrapper\\nodes_sampler.py\", line 3243, in process\n raise e\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\custom_nodes\\ComfyUI-WanVideoWrapper\\nodes_sampler.py\", line 2545, in process\n noise_pred, _, self.cache_state = predict_with_cfg(\n ^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\custom_nodes\\ComfyUI-WanVideoWrapper\\nodes_sampler.py\", line 1747, in predict_with_cfg\n raise e\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\custom_nodes\\ComfyUI-WanVideoWrapper\\nodes_sampler.py\", line 1594, in predict_with_cfg\n noise_pred_cond, noise_pred_ovi, cache_state_cond = transformer(\n ^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\nn\\modules\\module.py\", line 1775, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\nn\\modules\\module.py\", line 1786, in _call_impl\n return forward_call(*args, **kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\custom_nodes\\ComfyUI-WanVideoWrapper\\wanvideo\\modules\\model.py\", line 3173, in forward\n x, x_ip, lynx_ref_feature, x_ovi = block(x, x_ip=x_ip, lynx_ref_feature=lynx_ref_feature, x_ovi=x_ovi, x_onetoall_ref=x_onetoall_ref, onetoall_freqs=onetoall_freqs, **kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_dynamo\\eval_frame.py\", line 414, in __call__\n return super().__call__(*args, **kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\nn\\modules\\module.py\", line 1775, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\nn\\modules\\module.py\", line 1786, in _call_impl\n return forward_call(*args, **kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_dynamo\\eval_frame.py\", line 845, in compile_wrapper\n raise e.remove_dynamo_frames() from None # see TORCHDYNAMO_VERBOSE=1\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_inductor\\compile_fx.py\", line 990, in _compile_fx_inner\n raise InductorError(e, currentframe()).with_traceback(\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_inductor\\compile_fx.py\", line 974, in _compile_fx_inner\n mb_compiled_graph = fx_codegen_and_compile(\n ^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_inductor\\compile_fx.py\", line 1695, in fx_codegen_and_compile\n return scheme.codegen_and_compile(gm, example_inputs, inputs_to_check, graph_kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_inductor\\compile_fx.py\", line 1505, in codegen_and_compile\n compiled_module = graph.compile_to_module()\n ^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_inductor\\graph.py\", line 2319, in compile_to_module\n return self._compile_to_module()\n ^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_inductor\\graph.py\", line 2329, in _compile_to_module\n mod = self._compile_to_module_lines(wrapper_code)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_inductor\\graph.py\", line 2397, in _compile_to_module_lines\n mod = PyCodeCache.load_by_key_path(\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_inductor\\codecache.py\", line 3548, in load_by_key_path\n mod = _reload_python_module(key, path, set_sys_modules=in_toplevel)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_inductor\\runtime\\compile_tasks.py\", line 33, in _reload_python_module\n exec(code, mod.__dict__, mod.__dict__)\n File \"C:\\Users\\JARRA\\AppData\\Local\\Temp\\torchinductor_JARRA\\vt\\cvtcth3r2bq24xgslvskqy4lgsaspdjtwcxebaxkx4fd4q3tz5qx.py\", line 261, in \n triton_poi_fused___rshift___apply_lora_bitwise_and_bitwise_or_cat_lift_fresh_mul_select_split_split_with_sizes_sub_view_2 = async_compile.triton('triton_poi_fused___rshift___apply_lora_bitwise_and_bitwise_or_cat_lift_fresh_mul_select_split_split_with_sizes_sub_view_2', '''\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_inductor\\async_compile.py\", line 500, in triton\n kernel.precompile(\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py\", line 448, in precompile\n self._precompile_worker()\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py\", line 470, in _precompile_worker\n compile_results.append(self._precompile_config(c))\n ^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\torch\\_inductor\\runtime\\triton_heuristics.py\", line 778, in _precompile_config\n binary = triton.compile(*compile_args, **compile_kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\triton\\compiler\\compiler.py\", line 349, in compile\n fn_cache_manager.put_group(metadata_filename, metadata_group)\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\triton\\runtime\\cache.py\", line 96, in put_group\n return self.put(grp_contents, grp_filename, binary=False)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\JARRA\\Documents\\ComfyUI\\.venv\\Lib\\site-packages\\triton\\runtime\\cache.py\", line 117, in put\n with open(temp_path, mode) as f:\n ^^^^^^^^^^^^^^^^^^^^^\ntorch._inductor.exc.InductorError: FileNotFoundError: [Errno 2] No such file or directory: 'C:\\\\Users\\\\JARRA\\\\AppData\\\\Local\\\\Temp\\\\torchinductor_JARRA\\\\triton\\\\0\\\\YGMCYRRSOS4PIDJL6VS3N4XFTCS4MXREF32NRXHKMJAU7DJDQQNQ\\\\tmp.26fc6c94\\\\__grp__triton_poi_fused___rshift___apply_lora_bitwise_and_bitwise_or_cat_lift_fresh_mul_select_split_split_with_sizes_sub_view_2.json'\n\nSet TORCHDYNAMO_VERBOSE=1 for the internal stack trace (please do this especially if you're reporting a bug to PyTorch). For even more developer context, set TORCH_LOGS=\"+dynamo\"\n\n\nPrompt executed in 90.71 seconds\nSampling audio indices 0-81: 0%| | 0/6 [00:22 [17 lines of output]\n Traceback (most recent call last):\n File \"/opt/miniconda/envs/multitalk/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py\", line 389, in \n main()\n File \"/opt/miniconda/envs/multitalk/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py\", line 373, in main\n json_out[\"return_val\"] = hook(**hook_input[\"kwargs\"])\n File \"/opt/miniconda/envs/multitalk/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py\", line 143, in get_requires_for_build_wheel\n return hook(config_settings)\n File \"/tmp/pip-build-env-eqhjr_f7/overlay/lib/python3.10/site-packages/setuptools/build_meta.py\", line 331, in get_requires_for_build_wheel\n return self._get_build_requires(config_settings, requirements=[])\n File \"/tmp/pip-build-env-eqhjr_f7/overlay/lib/python3.10/site-packages/setuptools/build_meta.py\", line 301, in _get_build_requires\n self.run_setup()\n File \"/tmp/pip-build-env-eqhjr_f7/overlay/lib/python3.10/site-packages/setuptools/build_meta.py\", line 512, in run_setup\n super().run_setup(setup_script=setup_script)\n File \"/tmp/pip-build-env-eqhjr_f7/overlay/lib/python3.10/site-packages/setuptools/build_meta.py\", line 317, in run_setup\n exec(code, locals())\n File \"\", line 22, in \n ModuleNotFoundError: No module named 'torch'\n [end of output]\n \n note: This error originates from a subprocess, and is likely not a problem with pip.\nERROR: Failed to build 'flash_attn' when getting requirements to build wheel\n```\n\n```\n(multitalk) feixin@feixin:/data/deploy/InfiniteTalk$ pip list | grep torch\ntorch 2.4.1+cu121\ntorchaudio 2.4.1+cu121\ntorchvision 0.19.1+cu121\n```\n\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/177/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/176", "id": 3746955187, "node_id": "I_kwDOPdnatc7fVf-z", "number": 176, "title": "Run with the quantization model (Only support run with single gpu)", "user": { "login": "1390806607", "id": 89060957, "node_id": "MDQ6VXNlcjg5MDYwOTU3", "avatar_url": "https://avatars.githubusercontent.com/u/89060957?v=4", "gravatar_id": "", "url": "https://api.github.com/users/1390806607", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-12-19T10:21:20Z", "updated_at": "2025-12-19T10:21:20Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "When can quantitative models be used to support multi-card inference?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/176/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/175", "id": 3745922231, "node_id": "I_kwDOPdnatc7fRjy3", "number": 175, "title": "长视频生成会OOM", "user": { "login": "wriggleeeee", "id": 130350004, "node_id": "U_kgDOB8T7tA", "avatar_url": "https://avatars.githubusercontent.com/u/130350004?v=4", "gravatar_id": "", "url": "https://api.github.com/users/wriggleeeee", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-12-19T03:56:37Z", "updated_at": "2025-12-19T03:56:37Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "尝试生成190秒的视频，内存占用会一直增加，最后会OOM\n请问有大佬遇到这个问题吗？\nGPU：L20 显存48G，120G内存，用的LoRa，step是8", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/175/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/174", "id": 3745884521, "node_id": "I_kwDOPdnatc7fRalp", "number": 174, "title": "可能开源训练脚本嘛", "user": { "login": "gangxu822", "id": 24646369, "node_id": "MDQ6VXNlcjI0NjQ2MzY5", "avatar_url": "https://avatars.githubusercontent.com/u/24646369?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gangxu822", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-12-19T03:33:56Z", "updated_at": "2025-12-19T03:33:56Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": null, "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/174/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/173", "id": 3738756416, "node_id": "I_kwDOPdnatc7e2OVA", "number": 173, "title": "Floating point exception (core dumped)有人执行中遇到FPE问题吗", "user": { "login": "qingmeizhujiu", "id": 17784032, "node_id": "MDQ6VXNlcjE3Nzg0MDMy", "avatar_url": "https://avatars.githubusercontent.com/u/17784032?v=4", "gravatar_id": "", "url": "https://api.github.com/users/qingmeizhujiu", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-12-17T12:44:40Z", "updated_at": "2026-01-06T07:18:21Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "$python generate_infinitetalk.py --ckpt_dir weights/Wan2.1-I2V-14B-480P --wav2vec_dir 'weights/chinese-wav2vec2-base' --infinitetalk_dir weights/InfiniteTalk/single/infinitetalk.safetensors --input_json examples/single_example_image.json --size infinitetalk-720 --sample_steps 40 --mode streaming --motion_frame 9 --save_file infinitetalk_res_720p\n[2025-12-17 20:32:58,589] INFO: offload_model is not specified, set to True.\n[2025-12-17 20:32:58,589] INFO: Generation job args: Namespace(task='infinitetalk-14B', size='infinitetalk-720', frame_num=81, max_frame_num=1000, ckpt_dir='weights/Wan2.1-I2V-14B-480P', infinitetalk_dir='weights/InfiniteTalk/single/infinitetalk.safetensors', quant_dir=None, wav2vec_dir='weights/chinese-wav2vec2-base', dit_path=None, lora_dir=None, lora_scale=[1.2], offload_model=True, ulysses_size=1, ring_size=1, t5_fsdp=False, t5_cpu=False, dit_fsdp=False, save_file='infinitetalk_res_720p', audio_save_dir='save_audio', base_seed=42, input_json='examples/single_example_image.json', motion_frame=9, mode='streaming', sample_steps=40, sample_shift=11, sample_text_guide_scale=5.0, sample_audio_guide_scale=4.0, num_persistent_param_in_dit=None, audio_mode='localfile', use_teacache=False, teacache_thresh=0.2, use_apg=False, apg_momentum=-0.75, apg_norm_threshold=55, color_correction_strength=1.0, scene_seg=False, quant=None)\n[2025-12-17 20:32:58,589] INFO: Generation model config: {'__name__': 'Config: Wan MultiTalk AI2V 14B', 't5_model': 'umt5_xxl', 't5_dtype': torch.bfloat16, 'text_len': 512, 'param_dtype': torch.bfloat16, 'num_train_timesteps': 1000, 'sample_fps': 16, 'sample_neg_prompt': 'bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards', 't5_checkpoint': 'models_t5_umt5-xxl-enc-bf16.pth', 't5_tokenizer': 'google/umt5-xxl', 'clip_model': 'clip_xlm_roberta_vit_h_14', 'clip_dtype': torch.float16, 'clip_checkpoint': 'models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth', 'clip_tokenizer': 'xlm-roberta-large', 'vae_checkpoint': 'Wan2.1_VAE.pth', 'vae_stride': (4, 8, 8), 'patch_size': (1, 2, 2), 'dim': 5120, 'ffn_dim': 13824, 'freq_dim': 256, 'num_heads': 40, 'num_layers': 40, 'window_size': (-1, -1), 'qk_norm': True, 'cross_attn_norm': True, 'eps': 1e-06}\n[2025-12-17 20:32:58,589] INFO: Creating infinitetalk pipeline.\n[2025-12-17 20:32:58,589] INFO: loading weights/Wan2.1-I2V-14B-480P/models_t5_umt5-xxl-enc-bf16.pth\n[2025-12-17 20:34:02,253] INFO: loading weights/Wan2.1-I2V-14B-480P/Wan2.1_VAE.pth\n[2025-12-17 20:34:03,044] INFO: loading weights/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth\n[2025-12-17 20:34:08,147] INFO: Creating WanModel from weights/Wan2.1-I2V-14B-480P\n[2025-12-17 20:36:36,226] INFO: Generating video ...\n('/mnt/workspace/zhonggan/projects/talkingHead/InfiniteTalk-main/examples/single/ref_image.png', '/mnt/workspace/zhonggan/projects/talkingHead/InfiniteTalk-main/examples/single/1.wav')\nNo conversion needed.\n 0%| | 0/40 [00:00\n\nsingle: all FP32\n\"Image\"", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/165/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/164", "id": 3705497813, "node_id": "I_kwDOPdnatc7c3WjV", "number": 164, "title": "非官方微信社区 12.15过期", "user": { "login": "meiwulang", "id": 24241535, "node_id": "MDQ6VXNlcjI0MjQxNTM1", "avatar_url": "https://avatars.githubusercontent.com/u/24241535?v=4", "gravatar_id": "", "url": "https://api.github.com/users/meiwulang", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-12-08T09:33:04Z", "updated_at": "2025-12-08T09:33:04Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "![Image](https://github.com/user-attachments/assets/b72b9655-b52b-4a8f-bd3d-6cdee746dc2a)", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/164/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/163", "id": 3704553335, "node_id": "I_kwDOPdnatc7czv93", "number": 163, "title": "Run with FusioniX or Lightx2v(Require only 4~8 steps) with Wan2.1_I2V_14B_FusionX_LoRA do not work", "user": { "login": "yue001", "id": 4383606, "node_id": "MDQ6VXNlcjQzODM2MDY=", "avatar_url": "https://avatars.githubusercontent.com/u/4383606?v=4", "gravatar_id": "", "url": "https://api.github.com/users/yue001", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-12-08T04:05:52Z", "updated_at": "2026-01-07T19:02:53Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "As stated in the readme file, I ran the command using an f8 quantized model and FusioniX, but I found that the file was missing. I downloaded the file from [huggingface](https://huggingface.co/jrewingwannabe/Wan2.1_I2V_14B_FusionX_LoRA/) :\n\n```\npython generate_infinitetalk.py \\\n --ckpt_dir weights/Wan2.1-I2V-14B-480P \\\n --wav2vec_dir 'weights/chinese-wav2vec2-base' \\\n --infinitetalk_dir weights/InfiniteTalk/single/infinitetalk.safetensors \\\n --lora_dir weights/Wan2.1_I2V_14B_FusionX_LoRA/Wan2.1_I2V_14B_FusionX_LoRA.safetensors \\\n --input_json ./buen_input.json \\\n --lora_scale 1.0 \\\n --size infinitetalk-480 \\\n --sample_text_guide_scale 1.0 \\\n --sample_audio_guide_scale 2.0 \\\n --sample_steps 8 \\\n --mode streaming \\\n --quant fp8 \\\n --quant_dir weights/InfiniteTalk/quant_models/infinitetalk_single_fp8.safetensors \\\n --motion_frame 9 \\\n --sample_shift 2 \\\n --num_persistent_param_in_dit 0\n```\n\nbut the command logs did not show the Wan2.1_I2V_14B_FusionX_LoRA model being loaded, and there was no sign of improved inference speed. Has anyone successfully used this model to accelerate video generation?\n\n```\n[2025-12-08 11:46:47,855] INFO: offload_model is not specified, set to True.\n[2025-12-08 11:46:47,855] INFO: Generation job args: Namespace(task='infinitetalk-14B', size='infinitetalk-480', frame_num=81, max_frame_num=1000, ckpt_dir='weights/Wan2.1-I2V-14B-480P', infinitetalk_dir='weights/InfiniteTalk/single/infinitetalk.safetensors', quant_dir='weights/InfiniteTalk/quant_models/infinitetalk_single_fp8.safetensors', wav2vec_dir='weights/chinese-wav2vec2-base', dit_path=None, lora_dir=['weights/Wan2.1_I2V_14B_FusionX_LoRA/Wan2.1_I2V_14B_FusionX_LoRA.safetensors'], lora_scale=[1.0], offload_model=True, ulysses_size=1, ring_size=1, t5_fsdp=False, t5_cpu=False, dit_fsdp=False, save_file=None, audio_save_dir='save_audio', base_seed=42, input_json='./buen_input.json', motion_frame=9, mode='streaming', sample_steps=8, sample_shift=2.0, sample_text_guide_scale=1.0, sample_audio_guide_scale=2.0, num_persistent_param_in_dit=0, audio_mode='localfile', use_teacache=False, teacache_thresh=0.2, use_apg=False, apg_momentum=-0.75, apg_norm_threshold=55, color_correction_strength=1.0, scene_seg=False, quant='fp8')\n[2025-12-08 11:46:47,855] INFO: Generation model config: {'__name__': 'Config: Wan MultiTalk AI2V 14B', 't5_model': 'umt5_xxl', 't5_dtype': torch.bfloat16, 'text_len': 512, 'param_dtype': torch.bfloat16, 'num_train_timesteps': 1000, 'sample_fps': 16, 'sample_neg_prompt': 'bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards', 't5_checkpoint': 'models_t5_umt5-xxl-enc-bf16.pth', 't5_tokenizer': 'google/umt5-xxl', 'clip_model': 'clip_xlm_roberta_vit_h_14', 'clip_dtype': torch.float16, 'clip_checkpoint': 'models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth', 'clip_tokenizer': 'xlm-roberta-large', 'vae_checkpoint': 'Wan2.1_VAE.pth', 'vae_stride': (4, 8, 8), 'patch_size': (1, 2, 2), 'dim': 5120, 'ffn_dim': 13824, 'freq_dim': 256, 'num_heads': 40, 'num_layers': 40, 'window_size': (-1, -1), 'qk_norm': True, 'cross_attn_norm': True, 'eps': 1e-06}\n[2025-12-08 11:46:47,855] INFO: Creating infinitetalk pipeline.\n[2025-12-08 11:46:47,855] INFO: loading weights/Wan2.1-I2V-14B-480P/models_t5_umt5-xxl-enc-bf16.pth\n[2025-12-08 11:46:47,973] INFO: Loading quantized T5 from weights/InfiniteTalk/quant_models/t5_fp8.safetensors\n[2025-12-08 11:46:48,830] INFO: loading weights/Wan2.1-I2V-14B-480P/Wan2.1_VAE.pth\n[2025-12-08 11:46:49,326] INFO: loading weights/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth\n[2025-12-08 11:46:53,800] INFO: Creating WanModel from weights/Wan2.1-I2V-14B-480P\n[2025-12-08 11:46:53,800] INFO: Loading Quantized MultiTalk from weights/InfiniteTalk/quant_models/infinitetalk_single_fp8.safetensors\n[2025-12-08 11:46:55,660] INFO: Generating video ...\n('./buen_input/gril.png', './buen_input/pod_f_enhanced.wav')\n```\n\n\n\n\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/163/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/160", "id": 3663670980, "node_id": "I_kwDOPdnatc7aXy7E", "number": 160, "title": "关于显卡intel arc A770", "user": { "login": "jackson-lx", "id": 61775032, "node_id": "MDQ6VXNlcjYxNzc1MDMy", "avatar_url": "https://avatars.githubusercontent.com/u/61775032?v=4", "gravatar_id": "", "url": "https://api.github.com/users/jackson-lx", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-11-25T16:03:07Z", "updated_at": "2025-11-25T16:03:07Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "作者你好，在使用此代码本地部署运行，cpu是i9 11900kf+intel arc a770，只是出一个3秒的视频，都跑了1个小时仍然没有出来，原来的代码一直报错，使用ai修改了InfiniteTalk\\ComfyUI\\custom_nodes\\sageattention（新建的文件夹）\n在 sageattention 文件夹里新建两个文件（用记事本）：# Intel Arc A770 XPU 专用 SageAttention 兼容层（2025）文件 1：core.py\nimport torch\nimport torch.nn.functional as F\n\ndef sageattn(q, k, v, attn_mask=None, dropout_p=0, is_causal=False, tensor_layout=\"HND\"):\n print(\"[SageAttn XPU] 使用 PyTorch 原生 attention\")\n q = q.transpose(1, 2)\n k = k.transpose(1, 2)\n v = v.transpose(1, 2)\n if not (q.dtype == k.dtype == v.dtype):\n k = k.to(q.dtype)\n v = v.to(q.dtype)\n out = F.scaled_dot_product_attention(q, k, v, attn_mask=attn_mask, dropout_p=dropout_p, is_causal=is_causal)\n return out.transpose(1, 2).contiguous()\n\ndef sageattn_varlen(q, k, v, cu_seqlens_q, cu_seqlens_k, max_seqlen_q, max_seqlen_k, dropout_p=0, is_causal=False):\n print(\"[SageAttn Varlen XPU] 使用 PyTorch 原生 attention\")\n q = q.transpose(1, 2)\n k = k.transpose(1, 2)\n v = v.transpose(1, 2)\n out = F.scaled_dot_product_attention(q, k, v, dropout_p=dropout_p, is_causal=is_causal)\n return out.transpose(1, 2).contiguous()\n\ndef sageattn_blackwell(q, k, v, per_block_mean=False):\n print(\"[SageAttn Blackwell XPU] 使用 PyTorch 原生 attention\")\n q = q.transpose(1, 2)\n k = k.transpose(1, 2)\n v = v.transpose(1, 2)\n out = F.scaled_dot_product_attention(q, k, v)\n return out.transpose(1, 2).contiguous()\n\n__all__ = ['sageattn', 'sageattn_varlen', 'sageattn_blackwell']\n文件 2：__init__.py\nfrom .core import sageattn, sageattn_varlen, sageattn_blackwell\n\n# 伪造 CUDA 架构索引，彻底避免 NoneType\ndef get_cuda_arch_versions():\n return [80]\n\nimport torch\ntorch.cuda.get_device_properties = lambda device: type('obj', (object,), {'major': 8, 'minor': 0})()\n启动脚本\n@echo off\ncd /d \"%~dp0\"\nset CUDA_VISIBLE_DEVICES=\nset SYCL_CACHE_PERSISTENT=1\nset TORCH_XPU_VISIBLE_DEVICES=0\nset XPU_MEMORY_FRACTION=0.90\necho A770 8GB XPU 终极版 - 已完美兼容\npy312\\python.exe -s app.py --device xpu --lowvram --cpu-vae\npause\n终于是可以arc a770上跑起来，但是时间还是太长， 以下是ai的分析，仅供参考，请这个能修复吗？\n\n这意味着这个版本的 Infinite Talk 闭源程序 (cy_app.pyd) 根本不兼容您的 Arc A770 配置，其 Bug 无法通过环境变量或输入数据来绕过。\n\n在这种情况下，您唯一的解决路径是：\n\n联系 Infinite Talk 软件的作者或维护者，报告您在 Arc A770 (XPU) 上遇到的 TypeError 和 长时间挂起 的问题，针对 XPU 优化的补丁或更新的 pyd 文件。", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/160/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/159", "id": 3657885873, "node_id": "I_kwDOPdnatc7aBuix", "number": 159, "title": "RTX Pro 6000 (BlackWell sm_120) runs successfully, but the build process is extremely slow. Seeking diagnosis.", "user": { "login": "harleywang", "id": 2900178, "node_id": "MDQ6VXNlcjI5MDAxNzg=", "avatar_url": "https://avatars.githubusercontent.com/u/2900178?v=4", "gravatar_id": "", "url": "https://api.github.com/users/harleywang", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-11-24T09:15:37Z", "updated_at": "2025-12-04T08:05:14Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "My environment information is as follows:\n=============================================\n✅ SYSTEM INFORMATION\n=============================================\nOS: Linux 6.8.0-86-generic\nMachine: x86_64\nHostname: ubuntu\nCPU: x86_64\nPython: 3.11.13\n\n=============================================\n✅ PYTHON MODULES\n=============================================\nTorch: 2.9.1+cu129\nXformers: 0.0.32+8354497d.d20251117\nFlash Attn: 2.7.4.post1\n\n=============================================\n✅ TORCH CUDA INFORMATION\n=============================================\nCUDA Available: True\nCUDA Version from Torch: 12.9\nGPU Count: 1\nGPU 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition\n\n=============================================\n✅ SYSTEM CUDA INFORMATION (nvcc & nvidia-smi)\n=============================================\nnvcc: nvcc: NVIDIA (R) Cuda compiler driver\nCopyright (c) 2005-2021 NVIDIA Corporation\nBuilt on Thu_Nov_18_09:45:30_PST_2021\nCuda compilation tools, release 11.5, V11.5.119\nBuild cuda_11.5.r11.5/compiler.30672275_0\n\nnvidia-smi: Mon Nov 24 17:11:58 2025 \n+-----------------------------------------------------------------------------------------+\n| NVIDIA-SMI 580.95.05 Driver Version: 580.95.05 CUDA Version: 13.0 |\n+-----------------------------------------+------------------------+----------------------+\n| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |\n| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |\n| | | MIG M. |\n|=========================================+========================+======================|\n| 0 NVIDIA RTX PRO 6000 Blac... Off | 00000000:01:00.0 Off | Off |\n| 30% 31C P8 16W / 600W | 183MiB / 97887MiB | 0% Default |\n| | | N/A |\n+-----------------------------------------+------------------------+----------------------+\n\n+-----------------------------------------------------------------------------------------+\n| Processes: |\n| GPU GI CI PID Type Process name GPU Memory |\n| ID ID Usage |\n|=========================================================================================|\n| 0 N/A N/A 1457 G /usr/lib/xorg/Xorg 137MiB |\n| 0 N/A N/A 1893 G /usr/bin/gnome-shell 23MiB |\n+-----------------------------------------------------------------------------------------+\n\n\n\n\n\nBelow are screenshots after the video was run. An 11-second video ran for over 40 minutes, which is terrifyingly long:\n\n\"Image\"", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/159/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/158", "id": 3656153545, "node_id": "I_kwDOPdnatc7Z7HnJ", "number": 158, "title": "能否提供微调代码？", "user": { "login": "ILG2021", "id": 93691919, "node_id": "U_kgDOBZWgDw", "avatar_url": "https://avatars.githubusercontent.com/u/93691919?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ILG2021", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-11-23T15:41:27Z", "updated_at": "2025-11-23T15:41:27Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "需要lora微调以适应特殊风格", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/158/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/157", "id": 3656042507, "node_id": "I_kwDOPdnatc7Z6sgL", "number": 157, "title": "amd显卡不支持 有a卡版本吗！！求", "user": { "login": "beijitianhan-ops", "id": 236040257, "node_id": "U_kgDODhGwQQ", "avatar_url": "https://avatars.githubusercontent.com/u/236040257?v=4", "gravatar_id": "", "url": "https://api.github.com/users/beijitianhan-ops", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-11-23T14:04:33Z", "updated_at": "2025-11-23T14:04:33Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "amd显卡不支持 有a卡版本吗！！求", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/157/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/156", "id": 3655148400, "node_id": "I_kwDOPdnatc7Z3SNw", "number": 156, "title": "How to avoid messing up the source video? Limit infinitetalk to the face.", "user": { "login": "MiloMindbender", "id": 34351460, "node_id": "MDQ6VXNlcjM0MzUxNDYw", "avatar_url": "https://avatars.githubusercontent.com/u/34351460?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MiloMindbender", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-11-22T22:41:21Z", "updated_at": "2025-11-22T22:41:21Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "So I have done a bunch of videos with Wan 2.2 I'm using infinitetalk to apply lipsync to the character in the video.\n\nThe problem I've found is that since infinitetalk is based on Wan 2.1 it will sometimes do the face perfectly but will degrade motion or special effects in other parts of the video.\n\nIs there some way I can limit infinitetalk to just the area around the character's head? This needs to be done smoothly, so that there is not an obvious discontinuity or ripple around the head.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/156/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/155", "id": 3651331225, "node_id": "I_kwDOPdnatc7ZouSZ", "number": 155, "title": "How to load gguf model in gradio mode?", "user": { "login": "fahadh4ilyas", "id": 37577369, "node_id": "MDQ6VXNlcjM3NTc3MzY5", "avatar_url": "https://avatars.githubusercontent.com/u/37577369?v=4", "gravatar_id": "", "url": "https://api.github.com/users/fahadh4ilyas", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-11-21T11:38:16Z", "updated_at": "2025-11-21T11:38:16Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "I tried this model in A100 GPU and the speed is slower than my friend using comfyui. It seems that my friend is using gguf model to inference and I'm using full float16 model. When I tried using fp8 quant model, the speed is even slower than full model. I think gguf will be faster, but I can't load it using current main branch. Is there a way to load gguf? ", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/155/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/154", "id": 3645892318, "node_id": "I_kwDOPdnatc7ZT-be", "number": 154, "title": "RTX 5090D 不支持", "user": { "login": "qzw881130", "id": 3411500, "node_id": "MDQ6VXNlcjM0MTE1MDA=", "avatar_url": "https://avatars.githubusercontent.com/u/3411500?v=4", "gravatar_id": "", "url": "https://api.github.com/users/qzw881130", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-11-20T07:33:08Z", "updated_at": "2025-11-25T07:19:58Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "\"Image\"\n\n\"Image\"", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/154/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/151", "id": 3619822167, "node_id": "I_kwDOPdnatc7XwhpX", "number": 151, "title": "about the motion frame", "user": { "login": "tujie-jiangye", "id": 26537588, "node_id": "MDQ6VXNlcjI2NTM3NTg4", "avatar_url": "https://avatars.githubusercontent.com/u/26537588?v=4", "gravatar_id": "", "url": "https://api.github.com/users/tujie-jiangye", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-11-13T07:07:51Z", "updated_at": "2025-12-11T04:57:01Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "\"Image\"\n\n由于2处的存在，1处是不是多余的？", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/151/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/150", "id": 3615896613, "node_id": "I_kwDOPdnatc7XhjQl", "number": 150, "title": "Unable to Generate Full-Body Walking(e.g., TED Talk Style)", "user": { "login": "gamechanger-creator", "id": 222626544, "node_id": "U_kgDODUUC8A", "avatar_url": "https://avatars.githubusercontent.com/u/222626544?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gamechanger-creator", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-11-12T10:18:47Z", "updated_at": "2025-11-12T10:18:47Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hi team 👋,\n\nFirst of all, thank you for open-sourcing InfiniteTalk — it’s an incredible project with impressive lip-sync and expression alignment results!\n\nWhile testing the model, I noticed that most generated outputs (including the official examples) are limited to upper-body or half-body shots — usually up to the waist or chest level.\n\nIssue\n\nI’ve been trying to generate full-body, dynamic videos where a person is:\n\nStanding and walking on stage (like giving a TED Talk)\n\nMoving around while talking and addressing different sides of an audience\n\nHowever, even after several attempts, the generated video still remains half-body, and no movement like walking or turning is observed — even when I specify it clearly in the prompt or use different modes (streaming/clip) and model resolutions (480P, 720P).\n\nQuestions\n\nIs InfiniteTalk limited to upper-body motion because of its training data or architecture?\n\nAre there any configuration options or input formats (e.g., specific JSON fields or motion control parameters) that could help enable full-body motion?\n\nWould it be possible to extend the model or retrain it to support larger body movements and locomotion (e.g., walking, gesturing, turning)?\n\nIf InfiniteTalk is designed mainly for face and torso-level dubbing, could you please confirm that — so I can explore combining it with motion synthesis models for the lower body?\n\n\nTried prompts describing:\n\n“A person giving a TED talk, walking on stage, turning to the audience, using hand gestures.”\n\nBut the video still remains static beyond upper body movement.\n\nRequest\n\nCould you please guide me on how to generate such full-body, walking-talking style videos — or confirm if this limitation is due to the model design/training data?\n\nThank you again for your great work and open release! 🙏\nLooking forward to your guidance.\n\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/150/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/149", "id": 3608197737, "node_id": "I_kwDOPdnatc7XELpp", "number": 149, "title": "RUns this Exits with nothing Except: Creating WanModel from ./weights/Wan2.1-I2V-14B-480P", "user": { "login": "gjnave", "id": 59668562, "node_id": "MDQ6VXNlcjU5NjY4NTYy", "avatar_url": "https://avatars.githubusercontent.com/u/59668562?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gjnave", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-11-10T14:01:30Z", "updated_at": "2025-11-11T06:30:38Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "any ideas whats going on? it runs and says: Creating WanModel from ./weights/Wan2.1-I2V-14B-480P \nthen nothing.. whetehr i run one gpu or app", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/149/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/148", "id": 3596098486, "node_id": "I_kwDOPdnatc7WWBu2", "number": 148, "title": "运行gradio上传图片或音频都报502错，请问有人遇到吗？如何解决？", "user": { "login": "chrishjq", "id": 26054632, "node_id": "MDQ6VXNlcjI2MDU0NjMy", "avatar_url": "https://avatars.githubusercontent.com/u/26054632?v=4", "gravatar_id": "", "url": "https://api.github.com/users/chrishjq", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-11-06T14:38:03Z", "updated_at": "2025-11-06T14:38:03Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "\"Image\"\n\"Image\"\n\"Image\"\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/148/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/147", "id": 3594378063, "node_id": "I_kwDOPdnatc7WPdtP", "number": 147, "title": "Multi-GPU (A100 x2) Takes Longer per Iteration Than Single GPU", "user": { "login": "gamechanger-creator", "id": 222626544, "node_id": "U_kgDODUUC8A", "avatar_url": "https://avatars.githubusercontent.com/u/222626544?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gamechanger-creator", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 3, "created_at": "2025-11-06T06:52:47Z", "updated_at": "2026-01-05T10:19:03Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hello Team,\n\nI am facing a performance issue while running InfiniteTalk.\n\nHardware: A100 GPUs\n\nSingle GPU Mode: ~30 seconds per iteration\n\nDual GPU Mode (2× A100): ~120 seconds per iteration\n\nWhen I switch from 1 GPU to 2 GPUs, the time per iteration increases drastically, instead of decreasing. Ideally, multi-GPU should provide faster processing, but in this case, it becomes almost 4× slower.\n\nQuestions / Clarification Requested\n\nWhy does enabling multi-GPU increase the iteration time?\n\nIs there any specific configuration required for efficient multi-GPU usage?\n\nCould this be a DataParallel / DistributedDataParallel synchronization bottleneck?\n\nIs this expected behavior with the current InfiniteTalk implementation?\n\nExpected Behavior\n\nUsing more GPUs (2× A100) should reduce iteration time, not increase it.\nPlease let me know if there is a recommended method or configuration to achieve proper multi-GPU acceleration.\n\nThank you.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/147/reactions", "total_count": 1, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 1, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/146", "id": 3590664864, "node_id": "I_kwDOPdnatc7WBTKg", "number": 146, "title": "Face slightly changes at the end of the generated video when there is silence in the audio", "user": { "login": "gamechanger-creator", "id": 222626544, "node_id": "U_kgDODUUC8A", "avatar_url": "https://avatars.githubusercontent.com/u/222626544?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gamechanger-creator", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-11-05T11:54:22Z", "updated_at": "2025-11-05T11:54:22Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hi team,\n\nFirst of all, thank you for this amazing model — it’s generating really great and realistic videos.\n\nHowever, I’ve noticed an issue where the face changes slightly towards the end of the video, even though the same reference image is used throughout.\n\nIn my case, the input audio file has around 1 second of silence at the end of the dialogue. It seems the model is generating additional frames for that silent portion, which may be causing the face shift.\n\nI wanted to confirm:\n\nIs this behavior due to the trailing silence in the audio?\n\nOr is there any parameter or setting that can help maintain the same facial consistency (as in the starting frame) until the very end?\n\nAny guidance or fix for this would be greatly appreciated.\n\nThanks again for the excellent work!", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/146/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/145", "id": 3585788137, "node_id": "I_kwDOPdnatc7Vusjp", "number": 145, "title": "Support for Video-to-Video Generation in InfiniteTalkPipeline", "user": { "login": "asrays", "id": 71009078, "node_id": "MDQ6VXNlcjcxMDA5MDc4", "avatar_url": "https://avatars.githubusercontent.com/u/71009078?v=4", "gravatar_id": "", "url": "https://api.github.com/users/asrays", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-11-04T10:06:33Z", "updated_at": "2025-11-04T10:06:33Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Thank you for the excellent work on this project! \n\nAfter reviewing InfifniteTalk, I noticed that it’s mentioned to support video-to-video generation as well. However, in the generate_infinitetalk.py script, the wan.InfiniteTalkPipeline class seems to be initialized with components primarily designed for image-to-video generation.\n\nCould you please clarify how I can adapt or modify the pipeline for video-to-video generation? I might be missing something, so any guidance or pointers would be greatly appreciated.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/145/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/142", "id": 3570408665, "node_id": "I_kwDOPdnatc7U0BzZ", "number": 142, "title": "Would it be possible to replace the model wav2vec2-large-xlsr-53-chinese-zh-cn?", "user": { "login": "learitecnico", "id": 197441382, "node_id": "U_kgDOC8S3Zg", "avatar_url": "https://avatars.githubusercontent.com/u/197441382?v=4", "gravatar_id": "", "url": "https://api.github.com/users/learitecnico", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-10-30T11:49:21Z", "updated_at": "2025-11-04T00:48:51Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "I'm using the Infinite Talk for some video generation but i feel like some words are not being well pronounced in portuguese. I'm wondering if would be possible to replace the model wav2vec2-large-xlsr-53-chinese-zh-cn with another one, such as jonatasgrosman/wav2vec2-large-xlsr-53-portugues. If so, how can I do this and what changes could it imply for the workflow?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/142/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/140", "id": 3547740645, "node_id": "I_kwDOPdnatc7Tdjnl", "number": 140, "title": "Release Training Code", "user": { "login": "00Neil", "id": 76234364, "node_id": "MDQ6VXNlcjc2MjM0MzY0", "avatar_url": "https://avatars.githubusercontent.com/u/76234364?v=4", "gravatar_id": "", "url": "https://api.github.com/users/00Neil", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-10-24T05:40:25Z", "updated_at": "2025-10-24T05:40:25Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hello, thank you for the great work and for releasing the model! Could you please also consider releasing the code used for training this model?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/140/reactions", "total_count": 1, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 1 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/139", "id": 3545329305, "node_id": "I_kwDOPdnatc7TUW6Z", "number": 139, "title": "Ben 10 vc jacaré da favela", "user": { "login": "gustavogabirobap-ship-it", "id": 239832707, "node_id": "U_kgDODkuOgw", "avatar_url": "https://avatars.githubusercontent.com/u/239832707?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gustavogabirobap-ship-it", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-10-23T15:48:45Z", "updated_at": "2025-10-23T15:48:45Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "faça um vídeo na favela do Rio de Janeiro aonde o ben 10 aparece para pegar uma bicicleta que está escorada em uma das casas, para persegui um jacaré que está no meio da rua dando tiro pro alto, após o jacaré perceber isso ele logo começa tentar atirar no ben 10 porém o ben 10 pula em cima de uma casa e sai correndo Faça esse vídeo dentro de 50 segundos no tema minecraft ", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/139/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/138", "id": 3536000146, "node_id": "I_kwDOPdnatc7SwxSS", "number": 138, "title": "32G RAM+24G 5090+windows部署好后，模型运行不报错，直接退出了，使用的量化模型。", "user": { "login": "bomb-coin", "id": 87158576, "node_id": "MDQ6VXNlcjg3MTU4NTc2", "avatar_url": "https://avatars.githubusercontent.com/u/87158576?v=4", "gravatar_id": "", "url": "https://api.github.com/users/bomb-coin", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-10-21T10:55:15Z", "updated_at": "2025-10-21T10:55:15Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "python app.py --ckpt_dir weights/Wan2.1-I2V-14B-480P --wav2vec_dir 'weights/chinese-wav2vec2-base' --infinitetalk_dir weights/InfiniteTalk/single/infinitetalk.safetensors --quant fp8 --quant_dir weights/InfiniteTalk/quant_models/infinitetalk_single_fp8.safetensors --num_persistent_param_in_dit 0 --motion_frame 9 \n\n\"Image\"\n(D:\\code\\InfiniteTalk\\venv) PS D:\\code\\InfiniteTalk> pip list\nPackage Version\n-------------------------- ------------\naccelerate 1.10.1\naddict 2.4.0\naiofiles 24.1.0\naiohappyeyeballs 2.6.1\naiohttp 3.13.1\naiosignal 1.4.0\nannotated-types 0.7.0\nanyio 4.11.0\nasync-timeout 5.0.1\nattrs 25.4.0\naudioread 3.0.1\nbabel 2.17.0\nbeautifulsoup4 4.14.2\nblis 1.3.0\nBrotli 1.1.0\ncatalogue 2.0.10\ncertifi 2025.10.5\ncffi 2.0.0\ncharset-normalizer 3.4.4\nclick 8.2.1\ncloudpathlib 0.23.0\ncolorama 0.4.6\nconfection 0.1.5\ncontourpy 1.3.2\ncryptography 46.0.3\ncsvw 3.7.0\ncurated-tokenizers 0.0.9\ncurated-transformers 0.1.1\ncycler 0.12.1\ncymem 2.0.11\ndashscope 1.24.6\ndecorator 4.4.2\ndecord 0.6.0\ndiffusers 0.35.2\nDistVAE 0.0.0b5\ndlinfo 2.0.0\ndocopt 0.6.2\neasydict 1.13\neinops 0.8.1\nespeakng-loader 0.2.4\nexceptiongroup 1.3.0\nfastapi 0.119.1\nffmpy 0.6.3\nfilelock 3.19.1\nflash_attn 2.8.2\nfonttools 4.60.1\nfrozenlist 1.8.0\nfsspec 2025.9.0\nftfy 6.3.1\nfuture 1.0.0\ngradio 5.49.1\ngradio_client 1.13.3\ngroovy 0.1.2\nh11 0.16.0\nh2 4.3.0\nhpack 4.1.0\nhttpcore 1.0.9\nhttpx 0.28.1\nhuggingface-hub 0.35.3\nhyperframe 6.1.0\nidna 3.11\nimageio 2.37.0\nimageio-ffmpeg 0.6.0\nimportlib_metadata 8.7.0\nisodate 0.7.2\nJinja2 3.1.6\njoblib 1.5.2\njsonschema 4.25.1\njsonschema-specifications 2025.9.1\nkiwisolver 1.4.9\nlangcodes 3.5.0\nlanguage_data 1.3.0\nlanguage-tags 1.2.0\nlazy_loader 0.4\nlibrosa 0.11.0\nllvmlite 0.45.1\nloguru 0.7.3\nmarisa-trie 1.3.1\nmarkdown-it-py 4.0.0\nMarkupSafe 2.1.5\nmatplotlib 3.10.7\nmdurl 0.1.2\nmisaki 0.9.4\nmoviepy 1.0.3\nmpmath 1.3.0\nmsgpack 1.1.2\nmultidict 6.7.0\nmunkres 1.1.4\nmurmurhash 1.0.13\nnetworkx 3.3\nninja 1.13.0\nnum2words 0.5.14\nnumba 0.62.1\nnumpy 2.2.6\nopencv-python 4.11.0.86\noptimum-quanto 0.2.6\norjson 3.11.3\npackaging 25.0\npandas 2.3.3\nphonemizer 3.3.0\nphonemizer-fork 3.3.2\npillow 11.3.0\npip 25.2\nplatformdirs 4.5.0\npooch 1.8.2\npreshed 3.0.10\nproglog 0.1.12\npropcache 0.4.1\npsutil 7.1.1\npycparser 2.22\npydantic 2.11.10\npydantic_core 2.33.2\npydub 0.25.1\nPygments 2.19.2\npyloudnorm 0.1.1\npyparsing 3.2.5\nPySocks 1.7.1\npython-dateutil 2.9.0.post0\npython-multipart 0.0.20\npytz 2025.2\nPyYAML 6.0.3\nrdflib 7.2.1\nreferencing 0.37.0\nregex 2025.9.18\nrequests 2.32.5\nrfc3986 1.5.0\nrich 14.2.0\nrpds-py 0.27.1\nruff 0.14.1\nsafehttpx 0.1.6\nsafetensors 0.6.2\nscenedetect 0.6.7.1\nscikit-image 0.25.2\nscikit-learn 1.7.2\nscipy 1.15.2\nsegments 2.3.0\nsemantic-version 2.10.0\nsentencepiece 0.2.1\nsetuptools 80.9.0\nshellingham 1.5.4\nsix 1.17.0\nsmart_open 7.4.0\nsniffio 1.3.1\nsoundfile 0.13.1\nsoupsieve 2.8\nsoxr 1.0.0\nspacy 3.8.7\nspacy-curated-transformers 0.3.1\nspacy-legacy 3.0.12\nspacy-loggers 1.0.5\nsrsly 2.5.1\nstandard-aifc 3.13.0\nstandard-sunau 3.13.0\nstarlette 0.48.0\nsympy 1.14.0\ntermcolor 3.1.0\nthinc 8.3.6\nthreadpoolctl 3.6.0\ntifffile 2025.5.10\ntokenizers 0.22.1\ntomlkit 0.13.3\ntorch 2.8.0+cu128\ntorchaudio 2.8.0+cu128\ntorchvision 0.23.0+cu128\ntqdm 4.67.1\ntransformers 4.57.1\ntriton 3.0.0\ntyper 0.19.2\ntyping_extensions 4.15.0\ntyping-inspection 0.4.2\ntzdata 2025.2\nunicodedata2 16.0.0\nuritemplate 4.2.0\nurllib3 2.5.0\nuvicorn 0.38.0\nwasabi 1.1.3\nwcwidth 0.2.14\nweasel 0.4.1\nwebsocket-client 1.9.0\nwebsockets 15.0.1\nwheel 0.45.1\nwin_inet_pton 1.1.0\nwin32_setctime 1.2.0\nwrapt 2.0.0\nxformers 0.0.32.post2\nxfuser 0.4.4\nyarl 1.22.0\nyunchang 0.6.3.post1\nzipp 3.23.0\nzstandard 0.25.0", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/138/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/135", "id": 3533943369, "node_id": "I_kwDOPdnatc7So7JJ", "number": 135, "title": "Should we use this wrapper or wanvideo?", "user": { "login": "BestofthebestinAI", "id": 210918294, "node_id": "U_kgDODJJblg", "avatar_url": "https://avatars.githubusercontent.com/u/210918294?v=4", "gravatar_id": "", "url": "https://api.github.com/users/BestofthebestinAI", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-10-20T21:04:23Z", "updated_at": "2025-10-20T21:04:23Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hey\nFirst, I dont think I see this custom node in the MANAGER avalable to install?\n2) I think it was incorporated to wanvideo I think? Can we use just that or are we mising on stuff if we dont use this custom node?\n\nThanks", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/135/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/134", "id": 3529027071, "node_id": "I_kwDOPdnatc7SWK3_", "number": 134, "title": "int8 support", "user": { "login": "CreatorLLC", "id": 154020707, "node_id": "U_kgDOCS4rYw", "avatar_url": "https://avatars.githubusercontent.com/u/154020707?v=4", "gravatar_id": "", "url": "https://api.github.com/users/CreatorLLC", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-10-18T17:32:43Z", "updated_at": "2025-10-18T17:32:43Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Could you upload t5_int8.safetensors?\nFileNotFoundError: No such file or directory: weights/InfiniteTalk/quant_models/t5_int8.safetensors", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/134/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/133", "id": 3528057206, "node_id": "I_kwDOPdnatc7SSeF2", "number": 133, "title": "Example of generating a longer video - \"scene cut\"", "user": { "login": "MrEdwards007", "id": 116316872, "node_id": "U_kgDOBu7ayA", "avatar_url": "https://avatars.githubusercontent.com/u/116316872?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MrEdwards007", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-10-18T04:18:07Z", "updated_at": "2025-10-18T04:18:07Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "I don't have an issue but a question or request.\nDoes anyone have a workflow they can demonstrate for a long video?\nI'm reading that it is possible and I see a video but I am not understanding the video-to-video, the converting an image to video, or the \"scene cut\" described here.\n\n\n\"is I2V limited to 40 seconds?\" https://github.com/MeiGen-AI/InfiniteTalk/issues/10#issuecomment-3209476224\n\n\nIf there is a demonstration of this technique\\process, that would be so much better.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/133/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/132", "id": 3524433739, "node_id": "I_kwDOPdnatc7SEpdL", "number": 132, "title": "Tensorrt accelerate", "user": { "login": "BooHwang", "id": 114040640, "node_id": "U_kgDOBswfQA", "avatar_url": "https://avatars.githubusercontent.com/u/114040640?v=4", "gravatar_id": "", "url": "https://api.github.com/users/BooHwang", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 6, "created_at": "2025-10-17T04:15:33Z", "updated_at": "2026-01-04T08:19:10Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "I export wan2.1 model to tensorrt engine, the one step inference speed of wan2.1 on H100 was reduced from 11s to 4s.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/132/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/130", "id": 3512926544, "node_id": "I_kwDOPdnatc7RYwFQ", "number": 130, "title": "到底什么机器能跑通呀 4张4090 显存直接爆掉了", "user": { "login": "lixinuy", "id": 223837720, "node_id": "U_kgDODVd-GA", "avatar_url": "https://avatars.githubusercontent.com/u/223837720?v=4", "gravatar_id": "", "url": "https://api.github.com/users/lixinuy", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-10-14T08:33:16Z", "updated_at": "2025-10-14T08:33:16Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "到底什么机器能跑通呀 4张4090 显存直接爆掉了", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/130/reactions", "total_count": 4, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 3 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/129", "id": 3508703739, "node_id": "I_kwDOPdnatc7RIpH7", "number": 129, "title": "有没有人在910B上跑通了的？", "user": { "login": "LYA-Ansel", "id": 63083216, "node_id": "MDQ6VXNlcjYzMDgzMjE2", "avatar_url": "https://avatars.githubusercontent.com/u/63083216?v=4", "gravatar_id": "", "url": "https://api.github.com/users/LYA-Ansel", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-10-13T06:50:59Z", "updated_at": "2025-10-13T06:50:59Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "RT", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/129/reactions", "total_count": 1, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 1 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/128", "id": 3508643450, "node_id": "I_kwDOPdnatc7RIaZ6", "number": 128, "title": "RTX5090不支援", "user": { "login": "n59097", "id": 8072390, "node_id": "MDQ6VXNlcjgwNzIzOTA=", "avatar_url": "https://avatars.githubusercontent.com/u/8072390?v=4", "gravatar_id": "", "url": "https://api.github.com/users/n59097", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-10-13T06:23:06Z", "updated_at": "2025-10-13T06:23:06Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "RTX5090不支援", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/128/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/127", "id": 3495628461, "node_id": "I_kwDOPdnatc7QWw6t", "number": 127, "title": "Will there be an update for Wan2.2?", "user": { "login": "MehmetcanTozlu", "id": 65121799, "node_id": "MDQ6VXNlcjY1MTIxNzk5", "avatar_url": "https://avatars.githubusercontent.com/u/65121799?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MehmetcanTozlu", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 3, "created_at": "2025-10-08T14:13:24Z", "updated_at": "2025-11-22T16:05:49Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "I want to use with Wan2.2.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/127/reactions", "total_count": 3, "+1": 3, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/126", "id": 3492900084, "node_id": "I_kwDOPdnatc7QMWz0", "number": 126, "title": "Colega médico", "user": { "login": "draevelyntudisco-eng", "id": 236685473, "node_id": "U_kgDODhuIoQ", "avatar_url": "https://avatars.githubusercontent.com/u/236685473?v=4", "gravatar_id": "", "url": "https://api.github.com/users/draevelyntudisco-eng", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-10-07T20:17:55Z", "updated_at": "2025-10-07T20:17:55Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "crie um video com o seguinte prompt: (Portuguese dialogues in Brazil, no subtitles, only spoken audio. Must be lip-synced and realistic)\n\nCreate a hyper-realistic, IRL-style video scene, filmed like a raw handheld vlog. The setting is outside a small public health clinic (\"postinho de saúde\") in a Brazilian town. The morning light is soft, with motorbikes passing and people chatting in the background. The camera is handheld, slightly shaky, as if recorded by a colleague or nurse. The sound of distant street traffic mixes with birds chirping and someone coughing nearby.\n\nA doctor in a white coat, stethoscope around his neck, walks out the clinic door, stops in front of the camera, and looks directly into the lens with confidence. His tone is friendly, educational, and engaging, like he’s talking to other professionals on social media.\n\nThe doctor speaks in Brazilian Portuguese, confident and sincere:\n“Colega médico, prescrever em saúde mental não é mais dever do psiquiatra…”\n“Se você ainda acha isso, vai ficar pra trás!”\n“Foi por isso que eu preparei um guia em Prescrições em Saúde Mental pra você — médico generalista, da família, da linha de frente…”\n“Se ainda tem dificuldade em tratar depressão, ansiedade ou TDAH, relaxa — isso pode acabar hoje mesmo!”\n“Clica no link aí embaixo que o guia já tá disponível.”\n\n⸻\n\nFinal reminder: The audio must be in Brazilian Portuguese only, and the video must contain no subtitles or on-screen text.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/126/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/125", "id": 3481164914, "node_id": "I_kwDOPdnatc7Pflxy", "number": 125, "title": "Produtos", "user": { "login": "tiglearenato-netizen", "id": 235897313, "node_id": "U_kgDODg-B4Q", "avatar_url": "https://avatars.githubusercontent.com/u/235897313?v=4", "gravatar_id": "", "url": "https://api.github.com/users/tiglearenato-netizen", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-10-03T13:41:02Z", "updated_at": "2025-10-03T13:41:02Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "\"Galera, se liga nessa parada! Se você achava que casa inteligente era coisa de filme, o Echo Pop chegou pra provar o contrário. Sério, esse é o Smart Speaker mais bolado e compacto que a Amazon já fez!\nPrimeiro, o design: ele é compacto, mas não se engane. O som que sai desse bichinho é robusto e preenche o ambiente. Esquece aquele som abafado; a qualidade sonora aqui é nota dez para ouvir seu podcast, sua playlist favorita ou a rádio.\nMas o show de verdade é com a Alexa. É só mandar um 'Alexa, toca funk' ou 'Alexa, apaga a luz da sala' (se você tiver lâmpada inteligente, claro!) e pronto. A assistente é rápida, entende tudo e ainda te ajuda com o tempo, notícias ou a colocar um timer.\nÉ o jeito mais fácil e barato de começar a montar a sua casa conectada. Ele não ocupa espaço, é estiloso e te dá um poder de comando que vicia. Quer um upgrade sem furar o orçamento? O Echo Pop é a pedida! Manda bala no seu!\"", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/125/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/123", "id": 3462049686, "node_id": "I_kwDOPdnatc7OWq-W", "number": 123, "title": "Easier Download to correct folder for Missing Models, Nodes, and Workflows for templates on remote GPU runs.", "user": { "login": "eric4479", "id": 607831, "node_id": "MDQ6VXNlcjYwNzgzMQ==", "avatar_url": "https://avatars.githubusercontent.com/u/607831?v=4", "gravatar_id": "", "url": "https://api.github.com/users/eric4479", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-28T15:08:10Z", "updated_at": "2025-09-28T15:08:10Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "When running ComfyUI, InfiniteTalk, or other custom nodes that require certain models, nodes, or files, example media files to download to the correct folder under models, or others, even if running on Runpod, GCP, or locally. When I currently run this on Runpod, I click download model it tried to save locally.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/123/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/122", "id": 3460989445, "node_id": "I_kwDOPdnatc7OSoIF", "number": 122, "title": "对于5090显卡的支持需要怎么调整代码？", "user": { "login": "heyyyyou", "id": 32932178, "node_id": "MDQ6VXNlcjMyOTMyMTc4", "avatar_url": "https://avatars.githubusercontent.com/u/32932178?v=4", "gravatar_id": "", "url": "https://api.github.com/users/heyyyyou", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 3, "created_at": "2025-09-28T02:55:50Z", "updated_at": "2025-11-10T06:30:31Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "CUDA error (/__w/xformers/xformers/third_party/flash-attention/hopper/flash_fwd_launch_template.h:180): invalid argument\n\n主要是flash-atten对5090显卡无法支持使用。", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/122/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/121", "id": 3460963716, "node_id": "I_kwDOPdnatc7OSh2E", "number": 121, "title": "搞不定环境，能不能出个docker镜像呀，各位出个镜像救救孩子吧", "user": { "login": "CDSR-JY", "id": 41277372, "node_id": "MDQ6VXNlcjQxMjc3Mzcy", "avatar_url": "https://avatars.githubusercontent.com/u/41277372?v=4", "gravatar_id": "", "url": "https://api.github.com/users/CDSR-JY", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-09-28T02:13:59Z", "updated_at": "2025-12-08T06:22:15Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "按照README安装一遍，各种版本不兼容问题，启动时各种报错，搞不定了。求求各位出个镜像救救孩子吧。", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/121/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/120", "id": 3460009465, "node_id": "I_kwDOPdnatc7OO435", "number": 120, "title": "这里有更快更高清的数字人sdk，欢迎测试👏", "user": { "login": "oneCodeSuperman", "id": 147574712, "node_id": "U_kgDOCMvPuA", "avatar_url": "https://avatars.githubusercontent.com/u/147574712?v=4", "gravatar_id": "", "url": "https://api.github.com/users/oneCodeSuperman", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-27T08:57:53Z", "updated_at": "2025-09-27T08:57:53Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "https://github.com/oneCodeSuperman/LstmSync", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/120/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/118", "id": 3451974936, "node_id": "I_kwDOPdnatc7NwPUY", "number": 118, "title": "不同机器，环境版本audio emb不一样", "user": { "login": "zhangquanwei962", "id": 61458769, "node_id": "MDQ6VXNlcjYxNDU4NzY5", "avatar_url": "https://avatars.githubusercontent.com/u/61458769?v=4", "gravatar_id": "", "url": "https://api.github.com/users/zhangquanwei962", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-25T05:35:57Z", "updated_at": "2025-09-25T05:35:57Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "作者你好，我发现在不同的环境和显卡版本中，我提取到的audio emb会有差距，这个会很影响结果吗？\n比如下面\n```\ntensor([[[ 0.3360, 0.2743, 0.1058, ..., 0.1338, 0.1703, 0.1998],\n [ 0.4092, 0.0860, -0.3267, ..., 0.2698, 0.0648, 0.1479],\n [ 0.0522, 0.3074, -0.1485, ..., 0.0891, -0.2076, -0.0793],\n ...,\n [ 0.1871, 0.3271, 0.1834, ..., 0.0513, 0.2526, 0.0451],\n [ 0.0276, 0.3917, 0.2534, ..., 0.0384, 0.1995, 0.1240],\n [-0.0515, 0.0304, -0.0581, ..., -0.2275, 0.0173, -0.0590]],\n\n [[ 0.1892, 0.2932, 0.0223, ..., -0.0391, 0.0468, 0.1135],\n [ 0.1915, 0.1519, -0.3548, ..., 0.1473, 0.0028, 0.1744],\n [-0.1257, 0.3032, -0.3669, ..., 0.0438, -0.1621, -0.0412],\n ...,\n [ 0.1965, 0.3411, 0.1851, ..., 0.0457, 0.2501, 0.0103],\n [ 0.0423, 0.4006, 0.2531, ..., 0.0390, 0.1949, 0.1247],\n [-0.0438, 0.0348, -0.0753, ..., -0.2289, 0.0212, -0.0552]],\n\n [[-0.0695, 0.0758, -0.2953, ..., -0.0953, -0.3143, -0.0247],\n [-0.1537, 0.0810, -0.2333, ..., 0.0394, -0.0388, 0.2368],\n [-0.2326, 0.4914, -0.3418, ..., 0.2879, 0.0922, -0.0245],\n ...,\n [ 0.1431, 0.2861, 0.2316, ..., 0.1072, 0.1743, 0.0284],\n [ 0.0424, 0.3915, 0.2540, ..., 0.0329, 0.1621, 0.1276],\n [-0.0267, 0.0354, -0.0870, ..., -0.2399, 0.0193, -0.0583]],\n\n ...,\n\n [[-0.1611, 0.2584, -0.2998, ..., 0.2183, 0.0242, -0.2220],\n [ 0.0203, 0.1719, 0.0119, ..., 0.3144, 0.1404, 0.1222],\n [-0.0811, -0.0893, 0.1722, ..., 0.1725, 0.2086, -0.0058],\n ...,\n [-0.0957, 0.3421, 0.1778, ..., -0.0169, 0.2748, 0.2064],\n [-0.0942, 0.3644, 0.2175, ..., 0.1463, 0.4972, 0.5480],\n [-0.2950, -0.0821, -0.0558, ..., -0.0835, 0.0599, 0.1298]],\n\n [[-0.1193, 0.0600, -0.1389, ..., 0.3360, 0.0809, -0.2914],\n [ 0.1207, 0.0990, 0.0274, ..., 0.1450, 0.0816, -0.1099],\n [ 0.0702, 0.0581, 0.0108, ..., -0.0047, -0.1317, 0.0267],\n ...,\n [ 0.0621, 0.2104, 0.0469, ..., -0.0670, 0.1590, 0.1935],\n [-0.0420, 0.5247, 0.3748, ..., 0.0060, 0.3474, 0.3752],\n [-0.0295, -0.0047, -0.0966, ..., -0.3823, -0.0794, 0.1247]],\n\n [[ 0.1618, -0.0383, -0.1087, ..., 0.4226, -0.0722, -0.1287],\n [ 0.2479, -0.0618, -0.0263, ..., 0.1189, 0.0460, -0.0071],\n [ 0.2652, -0.0397, 0.0101, ..., 0.1016, -0.1899, 0.0864],\n ...,\n [ 0.1732, 0.1297, 0.1480, ..., 0.0979, 0.2292, 0.0110],\n [ 0.0213, 0.3708, 0.6358, ..., -0.0123, 0.2459, 0.2201],\n [-0.0690, -0.0316, -0.0042, ..., -0.4040, 0.0038, 0.1375]]])\ntensor([[[ 3.1519e-01, 2.9406e-01, 1.1547e-01, ..., 1.6433e-01,\n 1.3486e-01, 2.0159e-01],\n [ 3.7973e-01, 1.0880e-01, -3.2387e-01, ..., 2.8011e-01,\n 4.0300e-02, 1.5048e-01],\n [ 4.5627e-02, 3.1989e-01, -1.5601e-01, ..., 1.1766e-01,\n -2.3724e-01, -8.6098e-02],\n ...,\n [ 1.8996e-01, 3.2371e-01, 1.8317e-01, ..., 5.2144e-02,\n 2.5838e-01, 4.8192e-02],\n [ 2.7472e-02, 3.8986e-01, 2.5480e-01, ..., 3.9164e-02,\n 2.0136e-01, 1.2406e-01],\n [-5.1407e-02, 3.0639e-02, -6.0449e-02, ..., -2.2170e-01,\n 1.8119e-02, -5.5280e-02]],\n\n [[ 2.0725e-01, 2.6246e-01, 2.8467e-04, ..., -3.1249e-02,\n 4.7389e-02, 1.0200e-01],\n [ 1.7887e-01, 1.6302e-01, -3.5770e-01, ..., 1.5184e-01,\n -6.1282e-03, 1.5658e-01],\n [-1.2956e-01, 3.0548e-01, -3.8013e-01, ..., 4.0232e-02,\n -1.8006e-01, -4.8988e-02],\n ...,\n [ 2.0018e-01, 3.3312e-01, 1.8695e-01, ..., 4.8077e-02,\n 2.5216e-01, 1.5104e-02],\n [ 4.1745e-02, 3.9774e-01, 2.5539e-01, ..., 3.9256e-02,\n 1.9799e-01, 1.2325e-01],\n [-4.4092e-02, 3.4633e-02, -7.6886e-02, ..., -2.2390e-01,\n 2.1571e-02, -5.2313e-02]],\n\n [[-8.6154e-02, 7.0467e-02, -3.0331e-01, ..., -1.3018e-01,\n -3.2197e-01, -3.3422e-02],\n [-1.7790e-01, 7.5976e-02, -2.6204e-01, ..., 1.1702e-02,\n -4.3035e-02, 2.4582e-01],\n [-2.3418e-01, 4.8961e-01, -3.4283e-01, ..., 2.7643e-01,\n 9.2122e-02, -3.5267e-02],\n ...,\n [ 1.4403e-01, 2.8291e-01, 2.3397e-01, ..., 1.0966e-01,\n 1.7691e-01, 2.9013e-02],\n [ 4.1761e-02, 3.8962e-01, 2.5445e-01, ..., 3.3439e-02,\n 1.6272e-01, 1.2640e-01],\n [-2.6541e-02, 3.5729e-02, -8.9439e-02, ..., -2.3450e-01,\n 2.0073e-02, -5.5591e-02]],\n\n ...,\n\n [[-1.6236e-01, 2.5967e-01, -2.9905e-01, ..., 2.1992e-01,\n 2.3825e-02, -2.2154e-01],\n [ 1.9361e-02, 1.7274e-01, 1.1271e-02, ..., 3.1370e-01,\n 1.3959e-01, 1.2159e-01],\n [-8.1536e-02, -8.9640e-02, 1.7199e-01, ..., 1.7412e-01,\n 2.0879e-01, -5.4475e-03],\n ...,\n [-9.5941e-02, 3.3924e-01, 1.7002e-01, ..., -1.0751e-02,\n 2.7540e-01, 2.1081e-01],\n [-9.4996e-02, 3.6339e-01, 2.0956e-01, ..., 1.4948e-01,\n 5.0623e-01, 5.4825e-01],\n [-2.9836e-01, -8.4980e-02, -5.1893e-02, ..., -7.5370e-02,\n 6.8064e-02, 1.2874e-01]],\n\n [[-1.1892e-01, 5.9153e-02, -1.3830e-01, ..., 3.3685e-01,\n 8.1742e-02, -2.9118e-01],\n [ 1.2033e-01, 9.9296e-02, 2.8140e-02, ..., 1.4552e-01,\n 8.2084e-02, -1.0947e-01],\n [ 7.0559e-02, 5.8020e-02, 1.0193e-02, ..., -4.5394e-03,\n -1.3174e-01, 2.7679e-02],\n ...,\n [ 6.3078e-02, 2.0968e-01, 4.4784e-02, ..., -6.6299e-02,\n 1.5143e-01, 2.0062e-01],\n [-4.0928e-02, 5.2660e-01, 3.7435e-01, ..., 2.9442e-03,\n 3.4882e-01, 3.7469e-01],\n [-3.0716e-02, -7.1205e-03, -9.6835e-02, ..., -3.7464e-01,\n -7.9177e-02, 1.2316e-01]],\n\n [[ 1.6068e-01, -3.8772e-02, -1.0824e-01, ..., 4.2313e-01,\n -7.2551e-02, -1.2848e-01],\n [ 2.4752e-01, -6.1660e-02, -2.5794e-02, ..., 1.1828e-01,\n 4.5593e-02, -5.6444e-03],\n [ 2.6337e-01, -3.8990e-02, 9.4375e-03, ..., 1.0208e-01,\n -1.8877e-01, 8.7663e-02],\n ...,\n [ 1.7794e-01, 1.3057e-01, 1.4146e-01, ..., 1.0145e-01,\n 2.2168e-01, 1.5291e-02],\n [ 2.2589e-02, 3.6870e-01, 6.3120e-01, ..., -1.0616e-02,\n 2.4266e-01, 2.1791e-01],\n [-6.9499e-02, -3.1871e-02, 3.3225e-04, ..., -4.0706e-01,\n 4.5749e-03, 1.4140e-01]]])\n同形状: True\n最大差异: 2.462098e+00\n不一致\n平均差异: 3.949627e-03\n\n```\n\n这俩是同一个文件在不同的环境提取的，竟然数值不一样", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/118/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/117", "id": 3449348851, "node_id": "I_kwDOPdnatc7NmOLz", "number": 117, "title": "multiple prompts to guide", "user": { "login": "00Neil", "id": 76234364, "node_id": "MDQ6VXNlcjc2MjM0MzY0", "avatar_url": "https://avatars.githubusercontent.com/u/76234364?v=4", "gravatar_id": "", "url": "https://api.github.com/users/00Neil", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-09-24T13:02:14Z", "updated_at": "2025-11-04T03:18:15Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Thank you for the great work. I'm wondering if it's possible to use multi-line prompts to guide the specific actions of each window, and I mean implementing this feature in Python code rather than comfyui.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/117/reactions", "total_count": 2, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 2 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/115", "id": 3447235422, "node_id": "I_kwDOPdnatc7NeKNe", "number": 115, "title": "[BUG] Installation Fails on Windows 10 with PyTorch 2.5.1 due to xformers Incompatibility", "user": { "login": "aj689689", "id": 223002540, "node_id": "U_kgDODUq_rA", "avatar_url": "https://avatars.githubusercontent.com/u/223002540?v=4", "gravatar_id": "", "url": "https://api.github.com/users/aj689689", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-24T01:41:24Z", "updated_at": "2025-09-24T01:41:24Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hello Developers,\n\nI am having trouble running InfiniteTalk on a modern Windows 10 setup. After extensive troubleshooting, I have identified a core incompatibility issue with the `xformers` dependency.\n\n**Problem Summary:**\nThe script `generate_infinitetalk.py` has a hard dependency on `xformers`. However, the available pre-compiled version of `xformers` (`0.0.27.post2`) is incompatible with the latest versions of PyTorch (e.g., `2.5.1`) on Windows, causing a DLL error (\"Entry Point Not Found\"). This creates a situation where the application cannot run.\n\n**My Environment:**\n- **OS:** Windows 10 Pro x64\n- **GPU:** NVIDIA GeForce RTX 4070 (12GB)\n- **Conda:** Miniconda3 (conda version 25.7.0)\n- **Python:** 3.10\n- **PyTorch:** 2.5.1 (installed from `pytorch` channel with `pytorch-cuda=12.1`)\n- **CUDA Toolkit:** 12.1\n- **C++ Compiler:** MSVC 19.44.35217 (from Visual Studio 2022 Build Tools v17.10)\n\n**Errors Encountered:**\n\n1. When `xformers` is installed via `pip`, running any script that imports it results in a `python.exe - Entry Point Not Found` DLL error. The missing entry point is `?_singleton@UndefinedTensorlmpl@c10@@0U12@A` in `xformers\\_C.pyd`, which points to a mismatch with the `c10` library in PyTorch 2.5.1.\n\n2. If `xformers` is NOT installed, `generate_infinitetalk.py` fails immediately with `ModuleNotFoundError: No module named 'xformers'`, confirming it is a mandatory dependency and not an optional accelerator.\n\n3. Additionally, attempting to install `flash-attn` also fails during compilation with the error `unsupported Microsoft Visual Studio version!`, as CUDA 12.1 does not officially support the new MSVC v17.10 compiler.\n\n**Expected Behavior:**\nIdeally, the program should either:\na) Have a fallback mechanism to run without `xformers` (e.g., in a slower, more compatible mode).\nb) Or, have access to a version of `xformers` that is compatible with modern PyTorch versions (2.5.1+) on Windows.\n\nThank you for your hard work on this amazing project. I hope this detailed report helps in resolving the issue for users with similar modern environments.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/115/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/114", "id": 3445597053, "node_id": "I_kwDOPdnatc7NX6N9", "number": 114, "title": "Several issues regarding type conversion", "user": { "login": "tujie-jiangye", "id": 26537588, "node_id": "MDQ6VXNlcjI2NTM3NTg4", "avatar_url": "https://avatars.githubusercontent.com/u/26537588?v=4", "gravatar_id": "", "url": "https://api.github.com/users/tujie-jiangye", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-23T14:45:09Z", "updated_at": "2025-09-23T14:45:09Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "I found many type conversions of amp.autocast(dtype=torch.float32) in the code and would like to ask the author about the reason for this approach. I found that it is not quite consistent with the Settings of wan2.1. During the training process, is the training based on the float32 type or bf16.\n\nThanks!\n\n\n\n`\n\nclass WanAttentionBlock(nn.Module):\n\n def __init__(self,\n cross_attn_type,\n dim,\n ffn_dim,\n num_heads,\n window_size=(-1, -1),\n qk_norm=True,\n cross_attn_norm=False,\n eps=1e-6,\n output_dim=768,\n norm_input_visual=True,\n class_range=24,\n class_interval=4):\n ...\n \n\n def forward(\n self,\n x, \n e,\n seq_lens, \n grid_sizes,\n freqs,\n context,\n context_lens, \n audio_embedding=None, \n ref_target_masks=None, \n human_num=None,\n ):\n\n dtype = x.dtype\n assert e.dtype == torch.float32\n with amp.autocast(dtype=torch.float32):\n e = (self.modulation.to(e.device) + e).chunk(6, dim=1)\n assert e[0].dtype == torch.float32\n\n # self-attention\n y, x_ref_attn_map = self.self_attn(\n (self.norm1(x).float() * (1 + e[1]) + e[0]).type_as(x), seq_lens, grid_sizes,\n freqs, ref_target_masks=ref_target_masks)\n with amp.autocast(dtype=torch.float32):\n x = x + y * e[2]\n \n x = x.to(dtype)\n\n # cross-attention of text\n x = x + self.cross_attn(self.norm3(x), context, context_lens)\n\n # cross attn of audio\n x_a = self.audio_cross_attn(self.norm_x(x), encoder_hidden_states=audio_embedding,\n shape=grid_sizes[0], x_ref_attn_map=x_ref_attn_map, human_num=human_num)\n x = x + x_a\n\n y = self.ffn((self.norm2(x).float() * (1 + e[4]) + e[3]).to(dtype))\n with amp.autocast(dtype=torch.float32):\n x = x + y * e[5]\n\n\n x = x.to(dtype)\n\n return x\n\n`", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/114/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/112", "id": 3443905018, "node_id": "I_kwDOPdnatc7NRdH6", "number": 112, "title": "About the audio process", "user": { "login": "523997931", "id": 49772078, "node_id": "MDQ6VXNlcjQ5NzcyMDc4", "avatar_url": "https://avatars.githubusercontent.com/u/49772078?v=4", "gravatar_id": "", "url": "https://api.github.com/users/523997931", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-23T06:35:05Z", "updated_at": "2025-09-23T06:35:05Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hello,\nthanks for your great work firstly, I wonder to know in the development of this model? Did you use whisper as the audio feature extractor？This is trained for ASR, I think it maybe better than wav2vec,(unsupervised you know). I think this maybe more rubust for some noise in the audio.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/112/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/111", "id": 3439229332, "node_id": "I_kwDOPdnatc7M_nmU", "number": 111, "title": "How to prevent object from moving backward?", "user": { "login": "AiD-teng", "id": 186009102, "node_id": "U_kgDOCxZGDg", "avatar_url": "https://avatars.githubusercontent.com/u/186009102?v=4", "gravatar_id": "", "url": "https://api.github.com/users/AiD-teng", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-22T05:18:01Z", "updated_at": "2025-09-22T05:18:01Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "it may be relative to comfyui workflow but still want to know how to prevent object from moving backward when working on long video ( that needs the funcion like context window )", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/111/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/110", "id": 3437688260, "node_id": "I_kwDOPdnatc7M5vXE", "number": 110, "title": "安装的时候版本错误的看过来（个人安装的经验总结）", "user": { "login": "yanght1001", "id": 34299028, "node_id": "MDQ6VXNlcjM0Mjk5MDI4", "avatar_url": "https://avatars.githubusercontent.com/u/34299028?v=4", "gravatar_id": "", "url": "https://api.github.com/users/yanght1001", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 3, "created_at": "2025-09-21T01:21:52Z", "updated_at": "2025-11-04T07:37:29Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "1、必须安装在liunx下面，因为Windows下有的python包的版本不存在。\n2、requirements.txt 包的版本不精确，我把我调试后的版本发给大家。\n3、下面是具体的安装：\n操作系统版本：Ubuntu 24.04（Windows WSL2）\ncoda版本:12.8\n\n**创建虚拟环境（conda-env-lock.yml文件内容在下面）**\n`conda create -n multitalk --file conda-env-lock.yml`\n\n\n**激活环境**\n`conda activate multitalk`\n\n\n**安装 Python 层精确版本**\n`pip install -r requirements-lock.txt`\n\n\n**conda-env-lock.yml的文件内容：**\n# This file may be used to create an environment using:\n# conda create --name --file \n# platform: linux-64\n# created-by: conda 25.7.0\n@EXPLICIT\nhttps://repo.anaconda.com/pkgs/main/linux-64/_libgcc_mutex-0.1-main.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/ca-certificates-2025.9.9-h06a4308_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/ld_impl_linux-64-2.40-h12ee557_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/libgfortran5-13.2.0-ha4646dd_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/libstdcxx-ng-13.2.0-hc0a3c3a_7.conda\nhttps://repo.anaconda.com/pkgs/main/noarch/tzdata-2025b-h04d1e81_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/libgfortran-ng-13.2.0-h69a702a_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/libgomp-11.2.0-h1234567_1.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/_openmp_mutex-5.1-1_gnu.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/libgcc-ng-11.2.0-h1234567_1.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/aom-3.6.0-h6a678d5_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/bzip2-1.0.8-h5eee18b_6.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/dav1d-1.2.1-h5eee18b_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/expat-2.7.1-h6a678d5_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/fribidi-1.0.10-h36c2ea0_0.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/gmp-6.2.1-h58526e2_0.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/graphite2-1.3.14-h295c915_1.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/icu-73.1-h6a678d5_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/jpeg-9e-h166bdaf_1.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/lame-3.100-h7f98852_1001.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/lerc-4.0.0-h6a678d5_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/libbrotlicommon-1.0.9-h166bdaf_7.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/libdeflate-1.22-h5eee18b_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/libffi-3.4.4-h6a678d5_1.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/libiconv-1.17-h166bdaf_0.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/libogg-1.3.4-h7f98852_1.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/libopenblas-0.3.21-h043d6bf_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/libopus-1.3.1-h7f98852_1.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/libuuid-1.41.5-h5eee18b_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/libvpx-1.11.0-h9c3ff4c_3.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/libwebp-base-1.3.2-h5eee18b_1.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/libzlib-1.3.1-hb25bd0a_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/lz4-c-1.9.3-h9c3ff4c_1.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/ncurses-6.5-h7934f7d_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/nettle-3.6-he412f7d_0.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/openh264-2.1.1-h4ff587b_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/openssl-3.0.17-h5eee18b_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/pixman-0.46.4-h7934f7d_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/pthread-stubs-0.3-h0ce48e5_1.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/x264-1!161.3030-h7f98852_1.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/x265-3.5-h924138e_3.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/xorg-libxau-1.0.12-h9b100fa_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/xorg-libxdmcp-1.1.5-h9b100fa_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/xorg-xorgproto-2024.1-h5eee18b_1.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/xz-5.6.4-h5eee18b_1.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/gettext-0.19.8.1-h73d1719_1008.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/gnutls-3.6.13-h85f3911_1.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/libavif-1.1.1-h5eee18b_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/libblas-3.9.0-16_linux64_openblas.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/libbrotlidec-1.0.9-h166bdaf_7.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/libbrotlienc-1.0.9-h166bdaf_7.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/libllvm11-11.1.0-hf817b99_3.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/libvorbis-1.3.7-h9c3ff4c_0.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/libxcb-1.17.0-h9b100fa_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/readline-8.3-hc2a1206_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/zlib-1.3.1-hb25bd0a_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/zstd-1.5.2-h8a70e8d_1.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/brotli-bin-1.0.9-h166bdaf_7.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/libcblas-3.9.0-16_linux64_openblas.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/libflac-1.3.4-h27087fc_0.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/liblapack-3.9.0-16_linux64_openblas.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/libpng-1.6.39-h5eee18b_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/libtiff-4.7.0-hde9077f_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/libxml2-2.13.8-hfdd30dd_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/pcre2-10.42-hebb0a14_1.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/sqlite-3.50.2-hb25bd0a_1.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/xorg-libx11-1.8.12-h9b100fa_1.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/brotli-1.0.9-h166bdaf_7.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/freetype-2.13.3-h4a9f257_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/lcms2-2.16-h92b89f2_1.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/libglib-2.84.2-h37c7471_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/libsndfile-1.0.31-h9c3ff4c_1.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/openjpeg-2.5.2-h0d4d230_1.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/tk-8.6.15-h54e0aa7_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/xorg-libxext-1.3.6-h9b100fa_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/xorg-libxrender-0.9.12-h9b100fa_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/ffmpeg-4.4.0-h6987444_4.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/fontconfig-2.14.1-h55d465d_3.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/python-3.10.18-h1a3bd86_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/cairo-1.18.4-h44eff21_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/certifi-2025.8.3-pyhd8ed1ab_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/charset-normalizer-3.4.3-pyhd8ed1ab_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/cycler-0.12.1-pyhd8ed1ab_1.conda\nhttps://conda.anaconda.org/conda-forge/noarch/decorator-5.2.1-pyhd8ed1ab_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/hpack-4.1.0-pyhd8ed1ab_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/hyperframe-6.1.0-pyhd8ed1ab_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/idna-3.10-pyhd8ed1ab_1.conda\nhttps://conda.anaconda.org/conda-forge/noarch/munkres-1.1.4-pyhd8ed1ab_1.conda\nhttps://conda.anaconda.org/conda-forge/noarch/packaging-25.0-pyh29332c3_1.conda\nhttps://conda.anaconda.org/conda-forge/noarch/platformdirs-4.4.0-pyhcf101f3_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/pycparser-2.22-pyh29332c3_1.conda\nhttps://conda.anaconda.org/conda-forge/noarch/pyparsing-3.2.4-pyhcf101f3_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/pysocks-1.7.1-pyha55dd90_7.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/python_abi-3.10-2_cp310.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/setuptools-78.1.1-py310h06a4308_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/six-1.17.0-pyhe01879c_1.conda\nhttps://conda.anaconda.org/conda-forge/noarch/threadpoolctl-3.6.0-pyhecae5ae_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/wheel-0.45.1-py310h06a4308_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/zipp-3.23.0-pyhd8ed1ab_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/audioread-3.0.1-py310hff52083_2.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/brotli-python-1.0.9-py310hd8f1fbe_7.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/cffi-1.15.0-py310h0fdd8cc_0.tar.bz2\nhttps://conda.anaconda.org/conda-forge/noarch/h2-4.3.0-pyhcf101f3_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/harfbuzz-10.2.0-hdfddeaa_1.conda\nhttps://conda.anaconda.org/conda-forge/noarch/importlib_resources-6.5.2-pyhd8ed1ab_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/joblib-1.5.2-pyhd8ed1ab_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/kiwisolver-1.4.2-py310hbf28c38_1.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/noarch/pip-25.2-pyhc872135_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/python-dateutil-2.9.0.post0-pyhe01879c_2.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/scipy-1.8.1-py310h7612f91_0.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/unicodedata2-14.0.0-py310h5764c6d_1.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/fonttools-4.33.3-py310h5764c6d_0.tar.bz2\nhttps://conda.anaconda.org/conda-forge/noarch/importlib-resources-6.5.2-pyhd8ed1ab_0.conda\nhttps://repo.anaconda.com/pkgs/main/linux-64/pillow-11.3.0-py310hb1c3d2d_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/pysoundfile-0.11.0-pyhd8ed1ab_0.tar.bz2\nhttps://conda.anaconda.org/conda-forge/linux-64/scikit-learn-1.1.1-py310hffb9edd_0.tar.bz2\nhttps://repo.anaconda.com/pkgs/main/linux-64/zstandard-0.19.0-py310h5eee18b_0.conda\nhttps://conda.anaconda.org/conda-forge/linux-64/matplotlib-base-3.5.2-py310h5701ce4_0.tar.bz2\nhttps://conda.anaconda.org/conda-forge/noarch/resampy-0.4.3-pyhd8ed1ab_1.conda\nhttps://conda.anaconda.org/conda-forge/noarch/urllib3-2.5.0-pyhd8ed1ab_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/requests-2.32.5-pyhd8ed1ab_0.conda\nhttps://conda.anaconda.org/conda-forge/noarch/pooch-1.8.2-pyhd8ed1ab_3.conda\nhttps://conda.anaconda.org/conda-forge/noarch/librosa-0.9.2-pyhd8ed1ab_0.tar.bz2\n\n\n**requirements-lock.txt 文件内容：**\n\naccelerate==1.10.1\naddict==2.4.0\naiofiles==24.1.0\naiohappyeyeballs==2.6.1\naiohttp==3.12.15\naiosignal==1.4.0\nannotated-types==0.7.0\nanyio==4.10.0\nasync-timeout==5.0.1\nattrs==25.3.0\naudioread @ file:///home/conda/feedstock_root/build_artifacts/audioread_1725357455116/work\nbabel==2.17.0\nbeautifulsoup4==4.13.5\nblinker==1.9.0\nblis==1.3.0\nBrotli==1.1.0\nbuild==1.3.0\ncatalogue==2.0.10\ncertifi @ file:///home/conda/feedstock_root/build_artifacts/certifi_1754231422783/work/certifi\ncffi @ file:///home/conda/feedstock_root/build_artifacts/cffi_1636046050867/work\ncharset-normalizer @ file:///home/conda/feedstock_root/build_artifacts/charset-normalizer_1754767332901/work\nclick==8.3.0\ncloudpathlib==0.22.0\nconfection==0.1.5\ncryptography==46.0.1\ncsvw==3.6.0\ncurated-tokenizers==0.0.9\ncurated-transformers==0.1.1\ncycler @ file:///home/conda/feedstock_root/build_artifacts/cycler_1733332471406/work\ncymem==2.0.11\ndashscope==1.24.5\ndecorator==4.4.2\ndecord==0.6.0\ndiffusers==0.35.1\nDistVAE==0.0.0b5\ndlinfo==2.0.0\ndocopt==0.6.2\neasydict==1.13\neinops==0.8.1\nespeakng-loader==0.2.4\nexceptiongroup==1.3.0\nfastapi==0.116.2\nffmpy==0.6.1\nfilelock==3.13.1\nflash_attn==2.7.4.post1\nFlask==3.1.2\nfonttools @ file:///home/conda/feedstock_root/build_artifacts/fonttools_1651017726246/work\nfrozenlist==1.7.0\nfsspec==2024.6.1\nftfy==6.3.1\nfuture==1.0.0\ngradio==5.46.1\ngradio_client==1.13.1\ngroovy==0.1.2\nh11==0.16.0\nh2 @ file:///home/conda/feedstock_root/build_artifacts/bld/rattler-build_h2_1756364871/work\nhf-xet==1.1.10\nhpack @ file:///home/conda/feedstock_root/build_artifacts/hpack_1737618293087/work\nhttpcore==1.0.9\nhttpx==0.28.1\nhuggingface-hub==0.35.0\nhyperframe @ file:///home/conda/feedstock_root/build_artifacts/hyperframe_1737618333194/work\nidna @ file:///home/conda/feedstock_root/build_artifacts/idna_1733211830134/work\nimageio==2.37.0\nimageio-ffmpeg==0.6.0\nimportlib_metadata==8.7.0\nimportlib_resources @ file:///home/conda/feedstock_root/build_artifacts/importlib_resources_1736252299705/work\niniconfig==2.1.0\nisodate==0.7.2\nitsdangerous==2.2.0\nJinja2==3.1.4\njoblib @ file:///home/conda/feedstock_root/build_artifacts/joblib_1756321760188/work\njsonschema==4.25.1\njsonschema-specifications==2025.9.1\nkiwisolver @ file:///home/conda/feedstock_root/build_artifacts/kiwisolver_1648854394530/work\nlangcodes==3.5.0\nlanguage-tags==1.2.0\nlanguage_data==1.3.0\nlazy_loader==0.4\nlibrosa @ file:///home/conda/feedstock_root/build_artifacts/librosa_1656336149289/work\nllvmlite==0.42.0\nloguru==0.7.3\nmarisa-trie==1.3.1\nmarkdown-it-py==4.0.0\nMarkupSafe==2.1.5\nmatplotlib @ file:///home/conda/feedstock_root/build_artifacts/matplotlib-suite_1651609453931/work\nmdurl==0.1.2\nmisaki==0.9.4\nmoviepy==1.0.3\nmpmath==1.3.0\nmsgpack==1.1.1\nmultidict==6.6.4\nmunkres==1.1.4\nmurmurhash==1.0.13\nnetworkx==3.3\nninja==1.13.0\nnum2words==0.5.14\nnumba==0.59.1\nnumpy==1.26.4\nnvidia-cublas-cu12==12.1.3.1\nnvidia-cuda-cupti-cu12==12.1.105\nnvidia-cuda-nvrtc-cu12==12.1.105\nnvidia-cuda-runtime-cu12==12.1.105\nnvidia-cudnn-cu12==9.1.0.70\nnvidia-cufft-cu12==11.0.2.54\nnvidia-curand-cu12==10.3.2.106\nnvidia-cusolver-cu12==11.4.5.107\nnvidia-cusparse-cu12==12.1.0.106\nnvidia-nccl-cu12==2.20.5\nnvidia-nvjitlink-cu12==12.9.86\nnvidia-nvtx-cu12==12.1.105\nopencv-python==4.11.0.86\noptimum-quanto==0.2.6\norjson==3.11.3\npackaging @ file:///home/conda/feedstock_root/build_artifacts/bld/rattler-build_packaging_1745345660/work\npandas==2.3.2\nphonemizer-fork==3.3.2\npillow==11.0.0\npip @ file:///croot/pip_1756710021410/work\npip-tools==7.5.0\nplatformdirs @ file:///home/conda/feedstock_root/build_artifacts/bld/rattler-build_platformdirs_1756227402/work\npluggy==1.6.0\npooch @ file:///home/conda/feedstock_root/build_artifacts/pooch_1754941678315/work\npreshed==3.0.10\nproglog==0.1.12\npropcache==0.3.2\nprotobuf==6.32.1\npsutil==7.1.0\npycparser @ file:///home/conda/feedstock_root/build_artifacts/bld/rattler-build_pycparser_1733195786/work\npydantic==2.11.9\npydantic_core==2.33.2\npydub==0.25.1\nPygments==2.19.2\npyloudnorm==0.1.1\npyparsing @ file:///home/conda/feedstock_root/build_artifacts/bld/rattler-build_pyparsing_1757767060/work\npyproject_hooks==1.2.0\nPySocks @ file:///home/conda/feedstock_root/build_artifacts/pysocks_1733217236728/work\npytest==8.4.2\npython-dateutil @ file:///home/conda/feedstock_root/build_artifacts/bld/rattler-build_python-dateutil_1751104122/work\npython-multipart==0.0.20\npytz==2025.2\nPyYAML==6.0.2\nray==2.49.2\nrdflib==7.2.1\nreferencing==0.36.2\nregex==2025.9.18\nrequests @ file:///home/conda/feedstock_root/build_artifacts/requests_1755614211359/work\nresampy @ file:///home/conda/feedstock_root/build_artifacts/resampy_1735829929447/work\nrfc3986==1.5.0\nrich==14.1.0\nrpds-py==0.27.1\nruff==0.13.1\nsafehttpx==0.1.6\nsafetensors==0.6.2\nscenedetect==0.6.7\nscikit-image==0.25.2\nscikit-learn @ file:///home/conda/feedstock_root/build_artifacts/scikit-learn_1652976692230/work\nscipy==1.15.3\nsegments==2.3.0\nsemantic-version==2.10.0\nsentencepiece==0.2.1\nsetuptools==78.1.1\nshellingham==1.5.4\nsix @ file:///home/conda/feedstock_root/build_artifacts/bld/rattler-build_six_1753199211/work\nsmart_open==7.3.1\nsniffio==1.3.1\nsoundfile @ file:///home/conda/feedstock_root/build_artifacts/pysoundfile_1665058838115/work\nsoupsieve==2.8\nspacy==3.8.7\nspacy-curated-transformers==0.3.1\nspacy-legacy==3.0.12\nspacy-loggers==1.0.5\nsrsly==2.5.1\nstarlette==0.48.0\nsympy==1.13.3\ntermcolor==3.1.0\nthinc==8.3.6\nthreadpoolctl @ file:///home/conda/feedstock_root/build_artifacts/threadpoolctl_1741878222898/work\ntifffile==2025.5.10\ntokenizers==0.21.0\ntomli==2.2.1\ntomlkit==0.13.3\ntorch==2.4.1+cu121\ntorchaudio==2.4.1+cu121\ntorchvision==0.19.1+cu121\ntqdm==4.67.1\ntransformers==4.49.0\ntriton==3.0.0\ntyper==0.19.1\ntyping-inspection==0.4.1\ntyping_extensions==4.15.0\ntzdata==2025.2\nunicodedata2 @ file:///home/conda/feedstock_root/build_artifacts/unicodedata2_1649111910890/work\nuritemplate==4.2.0\nurllib3 @ file:///home/conda/feedstock_root/build_artifacts/urllib3_1750271362675/work\nuvicorn==0.36.0\nwasabi==1.1.3\nwcwidth==0.2.13\nweasel==0.4.1\nwebsocket-client==1.8.0\nwebsockets==15.0.1\nWerkzeug==3.1.3\nwheel==0.45.1\nwrapt==1.17.3\nxformers==0.0.28\nxfuser==0.4.1\nyarl==1.20.1\nyunchang==0.6.3.post1\nzipp @ file:///home/conda/feedstock_root/build_artifacts/zipp_1749421620841/work\nzstandard @ file:///croot/zstandard_1677013143055/work\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/110/reactions", "total_count": 6, "+1": 6, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/109", "id": 3433430825, "node_id": "I_kwDOPdnatc7Mpf8p", "number": 109, "title": "camera movement in I2V", "user": { "login": "4ld3v", "id": 4486875, "node_id": "MDQ6VXNlcjQ0ODY4NzU=", "avatar_url": "https://avatars.githubusercontent.com/u/4486875?v=4", "gravatar_id": "", "url": "https://api.github.com/users/4ld3v", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-19T08:33:14Z", "updated_at": "2025-09-19T08:33:14Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "For I2V: Is it possible to have any control about camera movement? Pan, Zoom In/Out etc", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/109/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/108", "id": 3432444822, "node_id": "I_kwDOPdnatc7MlvOW", "number": 108, "title": "mulher cantando pop", "user": { "login": "joaodepaulo-rgb", "id": 233186647, "node_id": "U_kgDODeYlVw", "avatar_url": "https://avatars.githubusercontent.com/u/233186647?v=4", "gravatar_id": "", "url": "https://api.github.com/users/joaodepaulo-rgb", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-19T02:11:39Z", "updated_at": "2025-09-19T02:11:39Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "![Image](https://github.com/user-attachments/assets/5478e8b1-96f3-4347-9675-6924ba7fdff9)", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/108/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/107", "id": 3429935895, "node_id": "I_kwDOPdnatc7McKsX", "number": 107, "title": "When the character is lip-syncing, can the expression be sad or angry?", "user": { "login": "pyhenger", "id": 211038429, "node_id": "U_kgDODJQw3Q", "avatar_url": "https://avatars.githubusercontent.com/u/211038429?v=4", "gravatar_id": "", "url": "https://api.github.com/users/pyhenger", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-09-18T11:29:57Z", "updated_at": "2025-09-18T14:18:58Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "When the character is lip-syncing, can the expression be sad or angry? Right now, no matter how I write the prompt, the character always turns out happy and smiling.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/107/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/106", "id": 3426734449, "node_id": "I_kwDOPdnatc7MP9Fx", "number": 106, "title": "LCM distillation support", "user": { "login": "tujie-jiangye", "id": 26537588, "node_id": "MDQ6VXNlcjI2NTM3NTg4", "avatar_url": "https://avatars.githubusercontent.com/u/26537588?v=4", "gravatar_id": "", "url": "https://api.github.com/users/tujie-jiangye", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-17T15:11:44Z", "updated_at": "2025-09-17T15:11:44Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Will the LCM distillation model be released and when will it be supported", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/106/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/105", "id": 3426500990, "node_id": "I_kwDOPdnatc7MPEF-", "number": 105, "title": "Tentative Date for the release of the 1.5B model", "user": { "login": "gamechanger-creator", "id": 222626544, "node_id": "U_kgDODUUC8A", "avatar_url": "https://avatars.githubusercontent.com/u/222626544?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gamechanger-creator", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-17T14:13:24Z", "updated_at": "2025-09-17T14:13:24Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "I would like to know, when the 1.5B model , will be released , which is useful for running it in the realtime.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/105/reactions", "total_count": 1, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 1, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/104", "id": 3425253976, "node_id": "I_kwDOPdnatc7MKTpY", "number": 104, "title": "How long the demo example usually takes to execute?", "user": { "login": "wriggleeeee", "id": 130350004, "node_id": "U_kgDOB8T7tA", "avatar_url": "https://avatars.githubusercontent.com/u/130350004?v=4", "gravatar_id": "", "url": "https://api.github.com/users/wriggleeeee", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-09-17T08:30:29Z", "updated_at": "2025-11-20T14:19:36Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "May I ask how long the demo example usually takes to execute? I've been running the sample for nearly two hours with no results yet. Is this duration normal? My hardware specs:\n\nGPU: NVIDIA H20 (96GB VRAM) with a single GPU \nCPU: 24-core\nRAM: 120GB\nDisk: 100GB\n\nHere is command:\npython generate_infinitetalk.py \\\n --ckpt_dir weights/Wan2.1-I2V-14B-480P \\\n --wav2vec_dir 'weights/chinese-wav2vec2-base' \\\n --infinitetalk_dir weights/InfiniteTalk/single/infinitetalk.safetensors \\\n --input_json examples/single_example_image.json \\\n --size infinitetalk-480 \\\n --sample_steps 40 \\\n --mode streaming \\\n --motion_frame 9 \\\n --save_file infinitetalk_res\n\n\"Image\"", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/104/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/103", "id": 3424796580, "node_id": "I_kwDOPdnatc7MIj-k", "number": 103, "title": "comfyui跑出来只有第一帧，后面黑屏", "user": { "login": "hmcmmmmm", "id": 24271691, "node_id": "MDQ6VXNlcjI0MjcxNjkx", "avatar_url": "https://avatars.githubusercontent.com/u/24271691?v=4", "gravatar_id": "", "url": "https://api.github.com/users/hmcmmmmm", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 5, "created_at": "2025-09-17T05:58:48Z", "updated_at": "2025-12-24T02:58:08Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "我在comfyui中用很多版本的工作流，在本地跑出来都只有第一帧图像，后面都是黑屏。\n工作流在云上跑没有问题，改了很多设定，用了不同的模型也还是这样。\n\n实在没有办法了，不知道是不是环境导致，但是wan2.1 都是能出的，加上infinite talk就黑屏了。\n\n有大佬遇到过这个问题吗？求助，\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/103/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/102", "id": 3424177368, "node_id": "I_kwDOPdnatc7MGMzY", "number": 102, "title": "When will the LCM version be released?", "user": { "login": "MathewLi9527", "id": 3665731, "node_id": "MDQ6VXNlcjM2NjU3MzE=", "avatar_url": "https://avatars.githubusercontent.com/u/3665731?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MathewLi9527", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-09-17T00:57:31Z", "updated_at": "2025-09-17T01:03:31Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": null, "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/102/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/101", "id": 3420172444, "node_id": "I_kwDOPdnatc7L27Cc", "number": 101, "title": "原始命令直接跑出现视频黑幕问题", "user": { "login": "su161021", "id": 85338500, "node_id": "MDQ6VXNlcjg1MzM4NTAw", "avatar_url": "https://avatars.githubusercontent.com/u/85338500?v=4", "gravatar_id": "", "url": "https://api.github.com/users/su161021", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 4, "created_at": "2025-09-16T03:20:29Z", "updated_at": "2025-09-16T03:32:54Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "我只修改了这一条命令，其他的没动\n\"Image\"\n\n执行命令\n\nGPU_NUM=5\ntorchrun --nproc_per_node=$GPU_NUM --standalone generate_infinitetalk.py \\\n --ckpt_dir weights/Wan2.1-I2V-14B-480P \\\n --wav2vec_dir 'weights/chinese-wav2vec2-base' \\\n --infinitetalk_dir weights/InfiniteTalk/single/infinitetalk.safetensors \\\n --dit_fsdp --t5_fsdp \\\n --ulysses_size=$GPU_NUM \\\n --input_json examples/single_example_image.json \\\n --size infinitetalk-480 \\\n --sample_steps 40 \\\n --mode streaming \\\n --motion_frame 9 \\\n --save_file infinitetalk_res_multigpu\n\n返回效果\n\nhttps://github.com/user-attachments/assets/4115196f-4049-428f-a3a3-d9d47fa7c38a\n\n大佬们有人遇到过这个问题吗", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/101/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/99", "id": 3416482009, "node_id": "I_kwDOPdnatc7Lo2DZ", "number": 99, "title": "频繁卡死", "user": { "login": "treebytehaha", "id": 101254775, "node_id": "U_kgDOBgkGdw", "avatar_url": "https://avatars.githubusercontent.com/u/101254775?v=4", "gravatar_id": "", "url": "https://api.github.com/users/treebytehaha", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-09-15T06:22:09Z", "updated_at": "2025-09-16T00:55:19Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "4090，32G内容，24G显存，以及一台32G内存，48G显存，后来换成64G内存，48G显存以及卡死，是怎么回事，步骤都是按照上面说的，运行的python generate_infinitetalk.py \\\n --ckpt_dir weights/Wan2.1-I2V-14B-480P \\\n --infinitetalk_dir weights/InfiniteTalk/single/infinitetalk.safetensors \\\n --input_json examples/single_example_image.json \\\n --size infinitetalk-480 \\\n --mode streaming，或者其他命令都一样被卡死，是为什么呢？感谢解答", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/99/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/98", "id": 3416169174, "node_id": "I_kwDOPdnatc7LnprW", "number": 98, "title": "The generated video duration is slightly longer than the audio duration", "user": { "login": "lmc710731910", "id": 52406180, "node_id": "MDQ6VXNlcjUyNDA2MTgw", "avatar_url": "https://avatars.githubusercontent.com/u/52406180?v=4", "gravatar_id": "", "url": "https://api.github.com/users/lmc710731910", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-09-15T03:51:06Z", "updated_at": "2025-09-16T01:00:13Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hello, may I ask a question? Why is the duration of the generated video always slightly longer than the audio when my audio is 9s? The generated video will end up with some extra character silence clips. I am using ComfyUI WanVideoWrapper\n\n生成的视频时长比音频时长多出一部分\n\n你好，请教个问题，为什么我音频是9s的，生成的视频时长总比音频时长要长一点呢，生成的视频最后会有多出一部分的人物静音片段。 我使用的是ComfyUI-WanVideoWrapper", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/98/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/96", "id": 3415845192, "node_id": "I_kwDOPdnatc7LmalI", "number": 96, "title": "Can this run on a consumer GPU or is this strictly for Enterprise level GPU", "user": { "login": "trashkollector", "id": 198321547, "node_id": "U_kgDOC9Iliw", "avatar_url": "https://avatars.githubusercontent.com/u/198321547?v=4", "gravatar_id": "", "url": "https://api.github.com/users/trashkollector", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 4, "created_at": "2025-09-15T00:44:14Z", "updated_at": "2026-01-12T19:59:50Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "How much VRAM is needed?\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/96/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/94", "id": 3405003735, "node_id": "I_kwDOPdnatc7K9DvX", "number": 94, "title": "can you dockerize this project", "user": { "login": "James-CodeX", "id": 124869546, "node_id": "U_kgDOB3Fbqg", "avatar_url": "https://avatars.githubusercontent.com/u/124869546?v=4", "gravatar_id": "", "url": "https://api.github.com/users/James-CodeX", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 4, "created_at": "2025-09-11T06:29:06Z", "updated_at": "2025-09-12T02:25:19Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": null, "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/94/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/93", "id": 3404607534, "node_id": "I_kwDOPdnatc7K7jAu", "number": 93, "title": "InfiniteTalk 480P Blank Audio + UniAnimate Test", "user": { "login": "Taiwan1912", "id": 116326112, "node_id": "U_kgDOBu7-4A", "avatar_url": "https://avatars.githubusercontent.com/u/116326112?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Taiwan1912", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-11T03:31:06Z", "updated_at": "2025-09-11T03:31:06Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "https://www.reddit.com/r/StableDiffusion/comments/1nd9xxy/infinitetalk_480p_blank_audio_unianimate_test/\n\nRTX 4090 48G Vram\n\nModel: wan2.1_i2v_480p_14B_bf16\n\nLora:\n\nlightx2v_I2V_14B_480p_cfg_step_distill_rank256_bf16\n\nUniAnimate-Wan2.1-14B-Lora-12000-fp16\n\nResolution: 480x832\n\nframes: 81 *9 / 625\n\nRendering time: 1 min 17s *9 = 15min\n\nSteps: 4\n\nBlock Swap: 14\n\nAudio CFG:1\n\nVram: 34 GB", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/93/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/92", "id": 3404536160, "node_id": "I_kwDOPdnatc7K7Rlg", "number": 92, "title": "Questions about the \"EXPERIMENT\" section of the paper", "user": { "login": "ZerRui", "id": 34739958, "node_id": "MDQ6VXNlcjM0NzM5OTU4", "avatar_url": "https://avatars.githubusercontent.com/u/34739958?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ZerRui", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-09-11T02:53:08Z", "updated_at": "2025-09-16T02:02:37Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "“”“”Test datasets and evaluation metrics To rigorously evaluate our method across diverse scenarios,\nwe utilize three benchmark datasets: HDTF Zhang et al. (2021) and CelebV-HQ Zhu et al. (2022)\n(emphasizing facial dynamics) alongside EMTD Rang Meng (2025) (incorporating full-body movements). Following established dubbing evaluation protocols Li et al. (2024); Fei et al. (2025), we\nconstruct a test set of 120 videos by randomly sampling 40 videos per dataset and permuting their\naudio channels (replacing original audio with mismatched tracks) to simulate real dubbing conditions.\nWe use our model to perform long sequence video dubbing at 480 × 480 resolution. The temporal\nlength of generated results is the frame number of the the dubbing video input.“”“\n\n\nThank you and your team for your outstanding work. I have some questions for you. Please advise. Thank you!\n1. As stated in your paper, you took 40 videos from each of the three datasets and scrambled their audio. So, when testing, did you use image-to-video generation or video-to-video generation? For image-to-video generation, did you use the first frame of each video?\n2. FID requires two datasets. In your Image2Video experiment, how were the two datasets designed? Could you explain the design using the HDTF FID dataset in Table 2 as an example?\n\nI look forward to your reply. Thank you!\n\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/92/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/91", "id": 3401862830, "node_id": "I_kwDOPdnatc7KxE6u", "number": 91, "title": "Question: Any plans to retrain with WAN 2.2", "user": { "login": "Passenger12138", "id": 47623700, "node_id": "MDQ6VXNlcjQ3NjIzNzAw", "avatar_url": "https://avatars.githubusercontent.com/u/47623700?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Passenger12138", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-09-10T10:19:40Z", "updated_at": "2025-10-10T16:56:53Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hello, thank you for sharing this great work on *InfiniteTalk: Audio-driven Video Generation for Sparse-Frame Video Dubbing*!\n\nI noticed that the current implementation and experiments are based on **WAN 2.1**. Since **WAN 2.2** has recently been released with improved performance on video generation, I was wondering:\n\n* Do you have any plans to retrain or adapt InfiniteTalk with WAN 2.2?\n* If not, do you think the current framework could be easily extended to leverage WAN 2.2?\n\nI believe this might further enhance the dubbing quality and temporal consistency.\nLooking forward to your thoughts, and thanks again for your contribution!\n\nBest regards,\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/91/reactions", "total_count": 5, "+1": 5, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/90", "id": 3395666900, "node_id": "I_kwDOPdnatc7KZcPU", "number": 90, "title": "infinitetalk remove the gesture in the video", "user": { "login": "sun-peach", "id": 14579257, "node_id": "MDQ6VXNlcjE0NTc5MjU3", "avatar_url": "https://avatars.githubusercontent.com/u/14579257?v=4", "gravatar_id": "", "url": "https://api.github.com/users/sun-peach", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-08T20:40:43Z", "updated_at": "2025-09-08T20:40:43Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hi, thanks for the great work.\n\nBut when I use the infinitetalk, I find it will remove the original gestures in the video. I have a 10 sec video with the figure waving hands during the first 2 sec. But after I apply infinitetalk to the video to generate 40 sec video. The waving is removed. I tried different prompts, but the removal is always there. Is there any way to force the system not to modify the video too much?\n\nThank you.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/90/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/89", "id": 3395116900, "node_id": "I_kwDOPdnatc7KXV9k", "number": 89, "title": "InfiniteTalk 480P Blank Audio NSFW Test~1min", "user": { "login": "Taiwan1912", "id": 116326112, "node_id": "U_kgDOBu7-4A", "avatar_url": "https://avatars.githubusercontent.com/u/116326112?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Taiwan1912", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-09-08T17:55:36Z", "updated_at": "2025-11-25T03:22:37Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "https://www.reddit.com/r/unstable_diffusion/comments/1nber74/infinitetalk_480p_nsfw_test1min/?utm_source=share&utm_medium=mweb3x&utm_name=mweb3xcss&utm_term=1&utm_content=share_button\n\nI used blank audio as input to generate the video and used LORA for testing\n\nRTX 4090 48G Vram\nModel: wan2.1_i2v_480p_14B_bf16\nLora: lightx2v_I2V_14B_480p_cfg_step_distill_rank256_bf16\nResolution: 832x480\nframes: 81 *22 / 1575\nRendering time: 1 min 20s *22 = 31min\nSteps: 4\nBlock Swap: 14\nAudio CFG:1\nVram: 34 GB", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/89/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/88", "id": 3394264081, "node_id": "I_kwDOPdnatc7KUFwR", "number": 88, "title": "Plan on wan 2.2 5b ?", "user": { "login": "StellarCheng", "id": 122878624, "node_id": "U_kgDOB1L6oA", "avatar_url": "https://avatars.githubusercontent.com/u/122878624?v=4", "gravatar_id": "", "url": "https://api.github.com/users/StellarCheng", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-09-08T13:34:56Z", "updated_at": "2025-09-09T00:55:03Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Since 5b model is much faster than 14b, and the quality is better than 1.3b model, do you have any plan to release the audio model trained on wan 2.2 5B model? ", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/88/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/87", "id": 3394211570, "node_id": "I_kwDOPdnatc7KT47y", "number": 87, "title": "teacache problem", "user": { "login": "MathewLi9527", "id": 3665731, "node_id": "MDQ6VXNlcjM2NjU3MzE=", "avatar_url": "https://avatars.githubusercontent.com/u/3665731?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MathewLi9527", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-08T13:22:11Z", "updated_at": "2025-09-08T13:22:11Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "The teacache has three branches, cond, drop_text, uncond. However, when math.isclose(text_guide_scale, 1.0) is True, the model is called only twice. Will this mismatch cause some potential problems??? Like less cache hit???", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/87/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/86", "id": 3392006622, "node_id": "I_kwDOPdnatc7KLene", "number": 86, "title": "ValueError: The output_attentions attribute is not supported when using the attn_implementation set to sdpa. Please set it to 'eager' instead.", "user": { "login": "Ai-Cyr", "id": 82373970, "node_id": "MDQ6VXNlcjgyMzczOTcw", "avatar_url": "https://avatars.githubusercontent.com/u/82373970?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Ai-Cyr", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 3, "created_at": "2025-09-07T21:06:18Z", "updated_at": "2025-09-22T08:06:26Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "**Description**\n\nI’m running on 4× H100 NVL GPUs, and I encounter the following error when using wav2vec2:\nValueError: The output_attentions attribute is not supported when using the attn_implementation set to sdpa. Please set it to 'eager' instead.\n\n**Steps to reproduce**\n\nRun wav2vec2 with attn_implementation=\"sdpa\"\nSet output_attentions=True\nObserve the error\n\n**Workaround**\n\nManually set output_attn = False in wav2vec2.py.\n\n**Problem**\n\nWhile this workaround avoids the error, it seems to slow down generation significantly.\n\n**Expected behavior**\n\nBeing able to use output_attentions with sdpa (or at least avoid slowdown when disabling it).\n\n**Environment**\n\nGPUs: 4× H100 NVL", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/86/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/85", "id": 3388104675, "node_id": "I_kwDOPdnatc7J8l_j", "number": 85, "title": "modifying the code for video to video dubbing for aligned video as input and take the all the frames for dubbing instead of sparse.", "user": { "login": "ashwani-ver", "id": 97355511, "node_id": "U_kgDOBc2G9w", "avatar_url": "https://avatars.githubusercontent.com/u/97355511?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ashwani-ver", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-05T16:16:02Z", "updated_at": "2025-09-05T16:16:02Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hi @yzhang2016 ,\n\nThanks for making your wonderful work public.\nI did the inference for video to video dubbing. What I saw in the output is that it did not use all the frames for dubbing. Though the sync was very good.\nBut, If i want to preserve the motion and action of the person and only want modify the lip region frame by frame and give the aligned video as input to the model instead of full resolution. Is there ant possibility to do that?\nIf I modify the code to pick all the frames for dubbing, will the model give same result?\n\nThanks in advance for your response.\nHoping to hear soon.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/85/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/84", "id": 3387944482, "node_id": "I_kwDOPdnatc7J7-4i", "number": 84, "title": "No matching distribution found for xformers==0.0.28", "user": { "login": "Prometheus-first", "id": 80702346, "node_id": "MDQ6VXNlcjgwNzAyMzQ2", "avatar_url": "https://avatars.githubusercontent.com/u/80702346?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Prometheus-first", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 3, "created_at": "2025-09-05T15:25:16Z", "updated_at": "2026-01-09T08:52:57Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "(multitalk) E:\\study\\InfiniteTalk-main>pip install -U xformers==0.0.28 --index-url https://download.pytorch.org/whl/cu121\nLooking in indexes: https://download.pytorch.org/whl/cu121\nERROR: Could not find a version that satisfies the requirement xformers==0.0.28 (from versions: 0.0.22.post4, 0.0.22.post7, 0.0.23, 0.0.23.post1, 0.0.24, 0.0.25, 0.0.25.post1, 0.0.26, 0.0.26.post1, 0.0.27, 0.0.27.post1, 0.0.27.post2)\nERROR: No matching distribution found for xformers==0.0.28\n\n这是为啥呢，按照readme里到xformers就报错了\nwin11系统，cuda12.9", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/84/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/82", "id": 3386997794, "node_id": "I_kwDOPdnatc7J4Xwi", "number": 82, "title": "How to install?", "user": { "login": "aswordok", "id": 12123396, "node_id": "MDQ6VXNlcjEyMTIzMzk2", "avatar_url": "https://avatars.githubusercontent.com/u/12123396?v=4", "gravatar_id": "", "url": "https://api.github.com/users/aswordok", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-09-05T10:06:53Z", "updated_at": "2025-09-05T10:26:53Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "ImageOrVideoUpload\nHow is this node installed?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/82/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/81", "id": 3386204609, "node_id": "I_kwDOPdnatc7J1WHB", "number": 81, "title": "视频对口型会改变视频内容，有什么办法避免吗", "user": { "login": "yuyue616", "id": 214791356, "node_id": "U_kgDODM10vA", "avatar_url": "https://avatars.githubusercontent.com/u/214791356?v=4", "gravatar_id": "", "url": "https://api.github.com/users/yuyue616", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-09-05T04:19:48Z", "updated_at": "2025-11-14T09:36:34Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "我用数字人对口型，数字人手中的产品直接变了。有什么能保证画面内容完全不变，只对口型吗？", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/81/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/80", "id": 3383638414, "node_id": "I_kwDOPdnatc7JrjmO", "number": 80, "title": "Do contador de histórias ao criador de memórias.", "user": { "login": "renatomariztvrh-wq", "id": 230368516, "node_id": "U_kgDODbslBA", "avatar_url": "https://avatars.githubusercontent.com/u/230368516?v=4", "gravatar_id": "", "url": "https://api.github.com/users/renatomariztvrh-wq", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-04T12:50:09Z", "updated_at": "2025-09-04T12:50:09Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Repórter cinematográfico cansado do jornalismo cotidiano homenageia sua mãe chamada carinhosamente lica e assim foca seu sonho em abrir um salão de esas exclusivo com estilo exclusivo Farm house e agora contando histórias mais leves afetivas. \n\n![Image](https://github.com/user-attachments/assets/30ddabbd-69d2-43b9-8a64-93cffc881d0a)", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/80/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/78", "id": 3382323973, "node_id": "I_kwDOPdnatc7JmisF", "number": 78, "title": "一开始CPU空间不够，--num_persistent_param_in_dit 0 \\去掉又显存OOM", "user": { "login": "treebytehaha", "id": 101254775, "node_id": "U_kgDOBgkGdw", "avatar_url": "https://avatars.githubusercontent.com/u/101254775?v=4", "gravatar_id": "", "url": "https://api.github.com/users/treebytehaha", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-04T06:21:18Z", "updated_at": "2025-09-04T06:21:18Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "📝 环境信息\n\n硬件环境\n\nGPU: NVIDIA GeForce RTX 4090 (显存 24 GB)\n\nDriver Version: 575.57.08\n\nCUDA Version (from nvidia-smi): 12.9\n\n软件环境\n\n操作系统: Linux (Ubuntu / 其他，请补充具体版本号，如 lsb_release -a)\n\nPython: 3.10 (Anaconda 环境 multitalk)\n\nPyTorch: 2.4.1+cu118 (torch.__version__)\n\nCUDA runtime: 11.8 (torch.version.cuda)\n\nnvcc: 12.0.76 (nvcc --version)\n\nTransformers: (贴一下 pip show transformers | grep Version)\n\nDiffusers: (贴一下 pip show diffusers | grep Version)\n\nflash-attn: 2.6.1 (如果安装了，否则写 None)\n\n其它依赖: optimum-quanto（从报错栈看你已经装了）\n\n模型与脚本\n\n仓库: [MeiGen-AI/InfiniteTalk](https://github.com/MeiGen-AI/InfiniteTalk?utm_source=chatgpt.com)\n\n命令:\n\npython generate_infinitetalk.py \\\n --ckpt_dir weights/Wan2.1-I2V-14B-480P \\\n --wav2vec_dir 'weights/chinese-wav2vec2-base' \\\n --infinitetalk_dir weights/InfiniteTalk/single/infinitetalk.safetensors \\\n --input_json examples/single_example_image.json \\\n --size infinitetalk-480 \\\n --sample_steps 40 \\\n --mode streaming \\\n --quant fp8 \\\n --quant_dir weights/InfiniteTalk/quant_models/infinitetalk_single_fp8.safetensors \\\n --motion_frame 9 \\\n --save_file infinitetalk_res_quant\n\n\n报错日志（关键部分）\n\ntorch.OutOfMemoryError: CUDA out of memory. Tried to allocate 26.00 MiB. \nGPU 0 has a total capacity of 23.51 GiB of which 18.38 MiB is free. \nProcess xxxx has 3.78 GiB memory in use. \nIncluding non-PyTorch memory, this process has 19.70 GiB memory in use.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/78/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/77", "id": 3379972253, "node_id": "I_kwDOPdnatc7Jdkid", "number": 77, "title": "Numpy Compatibility Issue", "user": { "login": "tensorlabs123", "id": 190712380, "node_id": "U_kgDOC14KPA", "avatar_url": "https://avatars.githubusercontent.com/u/190712380?v=4", "gravatar_id": "", "url": "https://api.github.com/users/tensorlabs123", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 9, "created_at": "2025-09-03T13:54:57Z", "updated_at": "2025-11-06T21:48:35Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Getting different conflicts on numpy.\n-> numba 0.55.1 requires numpy<1.22,>=1.18, but you have numpy 2.2.6 which is incompatible.\n-> scipy 1.8.1 requires numpy<1.25.0,>=1.17.3, but you have numpy 2.2.6 which is incompatible.\nAfter resolving this getting again numpy error:\n-> File \"_ckdtree.pyx\", line 1, in init scipy.spatial._ckdtree\nValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/77/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/76", "id": 3378626554, "node_id": "I_kwDOPdnatc7JYb_6", "number": 76, "title": "CSIM metric", "user": { "login": "dwanfun", "id": 129486402, "node_id": "U_kgDOB7fOQg", "avatar_url": "https://avatars.githubusercontent.com/u/129486402?v=4", "gravatar_id": "", "url": "https://api.github.com/users/dwanfun", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-03T07:15:46Z", "updated_at": "2025-09-03T07:15:46Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "请问论文中的CSIM指标是什么？图像走clip后的余弦相似度？", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/76/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/75", "id": 3376590315, "node_id": "I_kwDOPdnatc7JQq3r", "number": 75, "title": "blackwell / rtx 5000 support?", "user": { "login": "rubentorresbonet", "id": 4907670, "node_id": "MDQ6VXNlcjQ5MDc2NzA=", "avatar_url": "https://avatars.githubusercontent.com/u/4907670?v=4", "gravatar_id": "", "url": "https://api.github.com/users/rubentorresbonet", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-09-02T15:48:57Z", "updated_at": "2025-09-12T10:44:07Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "I see there are a lot of hard-coded versions in the install instructions, e.g. cuda 12.1\n\nBlackwell seems to need torch with cuda 12.8, which is the default torch for a while now.\n\nAnyone succeeded at running infinite talk with modern hardware? rtx 5090, rtx 6000 and the such?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/75/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/74", "id": 3375166498, "node_id": "I_kwDOPdnatc7JLPQi", "number": 74, "title": "怎么给视频配音？", "user": { "login": "gangxu822", "id": 24646369, "node_id": "MDQ6VXNlcjI0NjQ2MzY5", "avatar_url": "https://avatars.githubusercontent.com/u/24646369?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gangxu822", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-09-02T09:20:47Z", "updated_at": "2025-09-04T01:46:51Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": null, "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/74/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/73", "id": 3374724361, "node_id": "I_kwDOPdnatc7JJjUJ", "number": 73, "title": "with the simple example (1 picture iN) i got in OUT a cloud of points, it minds NOTHING ??!!", "user": { "login": "ABIA2024", "id": 179321600, "node_id": "U_kgDOCrA7AA", "avatar_url": "https://avatars.githubusercontent.com/u/179321600?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ABIA2024", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-09-02T07:11:14Z", "updated_at": "2025-09-05T16:10:00Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "\"Image\"", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/73/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/72", "id": 3374720050, "node_id": "I_kwDOPdnatc7JJiQy", "number": 72, "title": "i have load your example with simple picture , without change anything and in don't understand, i get a \"video combine\" without picture! Only one cload of color points in move.How can i fix it please?", "user": { "login": "ABIA2024", "id": 179321600, "node_id": "U_kgDOCrA7AA", "avatar_url": "https://avatars.githubusercontent.com/u/179321600?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ABIA2024", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-02T07:09:34Z", "updated_at": "2025-09-02T07:09:34Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "with no change from the example (single picture) i have got this VIDEO COMBINE: Nothing??§§\n\n\"Image\"", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/72/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/71", "id": 3374671570, "node_id": "I_kwDOPdnatc7JJWbS", "number": 71, "title": "720P 4min CFG 1 & CFG 3 Test", "user": { "login": "Taiwan1912", "id": 116326112, "node_id": "U_kgDOBu7-4A", "avatar_url": "https://avatars.githubusercontent.com/u/116326112?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Taiwan1912", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-09-02T06:51:58Z", "updated_at": "2025-09-02T06:54:19Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "https://www.reddit.com/r/StableDiffusion/comments/1n6aqgo/infinitetalk_720p_test4min_cfg1_cfg3/\n\nRTX 4090 48G Vram\n\nModel: wan2.1_i2v_720p_14B_fp16_scaled\nLora: lightx2v_I2V_14B_480p_cfg_step_distill_rank256_bf16\nResolution: 1280x720\nRendering time:\n\nCFG 1 = 4 min *97 / 6h 28min\n\nCFG 2 = 9 min *97 / 14h 33min\n\nframes: 81 *97 / 6975\nSteps: 4\nBlock Swap: 14\nVram: 44 GB\n--------------------------\n\nPrompt:\n\nA young woman, approximately 18 years old, with shoulder-length black hair, faces the camera. She wears a gentle, confident smile. Her eyes are bright and focused, looking forward. Soft lighting, a close-up of her upper body, and a slightly blurred background create a warm and professional portrait\n\nThis is a Japanese pop ballad performed by a female singer. The song has a beautiful melody and sincere emotions. The lyrics express the expectation and joy of love. The rhythm is slow and touching\n\nThe woman's lip movements in the video are perfectly synchronized with the Japanese voice, lyrics pronunciation, and tone in the audio, creating a natural and expressive lip-sync effect. The woman's facial expressions are adjusted to match the mood of the song, making her appear to be singing authentically. Slight head movements, eye movements, and natural body language are allowed to enhance the video's realism and liveliness, but any unnatural or exaggerated movements are avoided. The visual style, lighting, and high quality of the original image are maintained, with the background remaining stable or with only subtle changes in depth of field\n--------------------------", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/71/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/70", "id": 3374402241, "node_id": "I_kwDOPdnatc7JIUrB", "number": 70, "title": "How to avoid others lip synning or talking in the background??", "user": { "login": "AiD-teng", "id": 186009102, "node_id": "U_kgDOCxZGDg", "avatar_url": "https://avatars.githubusercontent.com/u/186009102?v=4", "gravatar_id": "", "url": "https://api.github.com/users/AiD-teng", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 9, "created_at": "2025-09-02T04:57:43Z", "updated_at": "2025-11-13T08:07:40Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hi, I counter an issue that I only want the main character to talk but people behind him in the bakcground aslo talk with lip syn. Any idea about how to solve the issue? thanks in advance!", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/70/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/67", "id": 3371799435, "node_id": "I_kwDOPdnatc7I-ZOL", "number": 67, "title": "Clarification on generation speed, repeated sampling steps, and guidance for real-time use", "user": { "login": "gamechanger-creator", "id": 222626544, "node_id": "U_kgDODUUC8A", "avatar_url": "https://avatars.githubusercontent.com/u/222626544?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gamechanger-creator", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-09-01T08:41:44Z", "updated_at": "2025-09-02T04:36:31Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hi team,\n\nFirst of all, thanks for open-sourcing InfiniteTalk — I’ve been experimenting with it and it’s very promising.\n\nI ran the following command after downloading the models from Hugging Face:\n\npython generate_infinitetalk.py \\\n --ckpt_dir weights/Wan2.1-I2V-14B-480P \\\n --wav2vec_dir weights/chinese-wav2vec2-base \\\n --infinitetalk_dir weights/InfiniteTalk/single/infinitetalk.safetensors \\\n --input_json examples/single_example_image.json \\\n --size infinitetalk-480 \\\n --sample_steps 8 \\\n --mode streaming \\\n --quant fp8 \\\n --quant_dir weights/InfiniteTalk/quant_models/infinitetalk_single_fp8.safetensors \\\n --motion_frame 9 \\\n --num_persistent_param_in_dit 0 \\\n --save_file infinitetalk_res_quant\n\nObservations:\n\nEach step takes about 5 minutes on my setup, and with --sample_steps 8 it runs 8/8 three separate times.\n\nEffectively, one sample ends up taking a very long time (tens of minutes).\n\nI would like to understand why the sampler is running 8/8 multiple times (three times in my case). Is this expected behavior, or am I mis-configuring something?\n\nQuestions:\n\nHow can I decrease the generation time? Is there a recommended configuration (e.g., adjusting motion_frame, sample_steps, or using a different quantization setting)?\n\nWhy does the sampler show 8/8 repeating three times? Is this related to multi-stage refinement or a bug?\n\nI’d like to explore real-time or near real-time use of InfiniteTalk. Are there ongoing efforts toward streaming optimization, and can I contribute in this direction? If yes, please point me to where in the codebase I should start.\n\nThanks in advance — happy to contribute with testing, code, or optimization work.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/67/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/66", "id": 3371480073, "node_id": "I_kwDOPdnatc7I9LQJ", "number": 66, "title": "Model fails to run on RTX 3050 (6GB VRAM + 8GB shared) – process stops after ‘Creating WanModel…", "user": { "login": "roossop24", "id": 217552217, "node_id": "U_kgDODPeVWQ", "avatar_url": "https://avatars.githubusercontent.com/u/217552217?v=4", "gravatar_id": "", "url": "https://api.github.com/users/roossop24", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-09-01T07:05:04Z", "updated_at": "2025-09-09T08:52:25Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "(venv) G:\\InfiniteTalk-main>python app.py\nA matching Triton is not available, some optimizations will not be enabled\nTraceback (most recent call last):\n File \"G:\\InfiniteTalk-main\\venv\\lib\\site-packages\\xformers\\__init__.py\", line 57, in _is_triton_available\n import triton # noqa\nModuleNotFoundError: No module named 'triton'\nWARNING 09-01 11:58:17 [envs.py:180] Flash Attention library \"flash_attn\" not found, using pytorch attention implementation\nW0901 11:58:18.073000 12872 venv\\Lib\\site-packages\\torch\\utils\\cpp_extension.py:2425] TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation.\nW0901 11:58:18.073000 12872 venv\\Lib\\site-packages\\torch\\utils\\cpp_extension.py:2425] If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'] to specific architectures.\n[2025-09-01 11:58:19,717] INFO: offload_model is not specified, set to True.\n[2025-09-01 11:58:19,717] INFO: Generation job args: Namespace(task='infinitetalk-14B', size='infinitetalk-480', frame_num=81, ckpt_dir='./weights/Wan2.1-I2V-14B-480P', quant_dir=None, infinitetalk_dir='weights/InfiniteTalk/single/infinitetalk.safetensors', wav2vec_dir='./weights/chinese-wav2vec2-base', dit_path=None, lora_dir=None, lora_scale=[1.2], offload_model=True, ulysses_size=1, ring_size=1, t5_fsdp=False, t5_cpu=False, dit_fsdp=False, save_file=None, audio_save_dir='save_audio/gradio', base_seed=42, input_json='examples.json', motion_frame=9, mode='streaming', sample_steps=40, sample_shift=7, sample_text_guide_scale=5.0, sample_audio_guide_scale=4.0, num_persistent_param_in_dit=None, use_teacache=False, teacache_thresh=0.2, use_apg=False, apg_momentum=-0.75, apg_norm_threshold=55, color_correction_strength=1.0, quant=None)\n[2025-09-01 11:58:19,717] INFO: Generation model config: {'__name__': 'Config: Wan MultiTalk AI2V 14B', 't5_model': 'umt5_xxl', 't5_dtype': torch.bfloat16, 'text_len': 512, 'param_dtype': torch.bfloat16, 'num_train_timesteps': 1000, 'sample_fps': 16, 'sample_neg_prompt': 'bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards', 't5_checkpoint': 'models_t5_umt5-xxl-enc-bf16.pth', 't5_tokenizer': 'google/umt5-xxl', 'clip_model': 'clip_xlm_roberta_vit_h_14', 'clip_dtype': torch.float16, 'clip_checkpoint': 'models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth', 'clip_tokenizer': 'xlm-roberta-large', 'vae_checkpoint': 'Wan2.1_VAE.pth', 'vae_stride': (4, 8, 8), 'patch_size': (1, 2, 2), 'dim': 5120, 'ffn_dim': 13824, 'freq_dim': 256, 'num_heads': 40, 'num_layers': 40, 'window_size': (-1, -1), 'qk_norm': True, 'cross_attn_norm': True, 'eps': 1e-06}\n[2025-09-01 11:58:20,633] INFO: Creating MultiTalk pipeline.\n[2025-09-01 11:58:20,644] INFO: loading ./weights/Wan2.1-I2V-14B-480P\\models_t5_umt5-xxl-enc-bf16.pth\n[2025-09-01 12:19:23,433] INFO: loading ./weights/Wan2.1-I2V-14B-480P\\models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth\n[2025-09-01 12:21:16,518] INFO: Creating WanModel from ./weights/Wan2.1-I2V-14B-480P\n\n\n### What I expected\nThe model should finish initializing or at least throw a CUDA OOM / error.\n\n### What happened\nProcess terminates right after \"Creating WanModel…\" without an error message.\n\n### Notes\n- RTX 3050 only has 6 GB VRAM. \n--offload_model True` is already set. \n- Tried lowering frame_num/sample_steps. \n\n### Request\n- Could you confirm the minimum VRAM required for the 14B pipeline? \n- Are smaller or quantized checkpoints available (7B/4B)? \n- Can the code fail fast with a clearer OOM message instead of silent exit? \n\nHappy to test with any debug flags you suggest.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/66/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/65", "id": 3370827448, "node_id": "I_kwDOPdnatc7I6r64", "number": 65, "title": "movement looks too exagerated and fake", "user": { "login": "trashkollector", "id": 198321547, "node_id": "U_kgDOC9Iliw", "avatar_url": "https://avatars.githubusercontent.com/u/198321547?v=4", "gravatar_id": "", "url": "https://api.github.com/users/trashkollector", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 8, "created_at": "2025-09-01T00:24:54Z", "updated_at": "2025-09-12T02:52:37Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Multitalk's movement is more realistic. \n\nWhy is infinite talk so exaggerated. can the movement be toned down so it is not so exaggerated.\n\nThe movement is very jerky at times. Any help would be greatly appreciated.\n\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/65/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/64", "id": 3369927879, "node_id": "I_kwDOPdnatc7I3QTH", "number": 64, "title": "How come it takes too much time to generate a 40s video?", "user": { "login": "baguiosfinest", "id": 28800292, "node_id": "MDQ6VXNlcjI4ODAwMjky", "avatar_url": "https://avatars.githubusercontent.com/u/28800292?v=4", "gravatar_id": "", "url": "https://api.github.com/users/baguiosfinest", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-31T07:16:12Z", "updated_at": "2025-09-02T05:06:30Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "I'm using rtx 3090 with 96gb RAM, Windows Linux.\n\nI have waited for almost 3 hours but still not complete so I just ended it. \nI also used the lowvram flag as the 480 and 720p crashed with CUDA error. \n\nDid I miss something? ", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/64/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/62", "id": 3369174790, "node_id": "I_kwDOPdnatc7I0YcG", "number": 62, "title": "error?", "user": { "login": "chef0531", "id": 195848012, "node_id": "U_kgDOC6xnTA", "avatar_url": "https://avatars.githubusercontent.com/u/195848012?v=4", "gravatar_id": "", "url": "https://api.github.com/users/chef0531", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-08-30T14:27:26Z", "updated_at": "2025-08-30T14:27:26Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Multitalk mode: infinitetalk\nSampling 288 frames in 5 windows, at 640x640 with 6 steps\nError during sampling: Sizes of tensors must match except in dimension 0. Expected size 80 but got size 40 for tensor number 1 in the list.\n!!! Exception during processing !!! Sizes of tensors must match except in dimension 0. Expected size 80 but got size 40 for tensor number 1 in the list.\nTraceback (most recent call last):\n File \"C:\\Users\\overn\\Desktop\\ComfyUI_windows_portable torch 2.7 cu128\\ComfyUI\\execution.py\", line 496, in execute\n output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, hidden_inputs=hidden_inputs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\overn\\Desktop\\ComfyUI_windows_portable torch 2.7 cu128\\ComfyUI\\execution.py\", line 315, in get_output_data\n return_values = await _async_map_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, hidden_inputs=hidden_inputs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"C:\\Users\\overn\\Desktop\\ComfyUI_windows_portable torch 2.7 cu128\\ComfyUI\\execution.py\", line 289, in _async_map_node_over_list\n await process_inputs(input_dict, i)\n File \"C:\\Users\\overn\\Desktop\\ComfyUI_windows_portable torch 2.7 cu128\\ComfyUI\\execution.py\", line 277, in process_inputs\n result = f(**inputs)\n ^^^^^^^^^^^\n File \"C:\\Users\\overn\\Desktop\\ComfyUI_windows_portable torch 2.7 cu128\\ComfyUI\\custom_nodes\\ComfyUI-WanVideoWrapper\\nodes.py\", line 3568, in process\n raise e\n File \"C:\\Users\\overn\\Desktop\\ComfyUI_windows_portable torch 2.7 cu128\\ComfyUI\\custom_nodes\\ComfyUI-WanVideoWrapper\\nodes.py\", line 3199, in process\n y = torch.cat([msk, y]) # 4+C T H W\n ^^^^^^^^^^^^^^^^^^^\nRuntimeError: Sizes of tensors must match except in dimension 0. Expected size 80 but got size 40 for tensor number 1 in the list.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/62/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/60", "id": 3367410357, "node_id": "I_kwDOPdnatc7Itpq1", "number": 60, "title": "can we use Wan 2.2 instead of Wan 2.1?", "user": { "login": "krigeta", "id": 75309361, "node_id": "MDQ6VXNlcjc1MzA5MzYx", "avatar_url": "https://avatars.githubusercontent.com/u/75309361?v=4", "gravatar_id": "", "url": "https://api.github.com/users/krigeta", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-08-29T16:23:13Z", "updated_at": "2025-08-29T16:23:13Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Wan 2.2 is far better than 2.1, can we use it with Infinitetalk?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/60/reactions", "total_count": 4, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 4 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/59", "id": 3366266914, "node_id": "I_kwDOPdnatc7IpSgi", "number": 59, "title": "Training details", "user": { "login": "HeathHose", "id": 14328405, "node_id": "MDQ6VXNlcjE0MzI4NDA1", "avatar_url": "https://avatars.githubusercontent.com/u/14328405?v=4", "gravatar_id": "", "url": "https://api.github.com/users/HeathHose", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-08-29T10:00:06Z", "updated_at": "2025-08-29T10:00:06Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hello, I tried to reproduce infinite. But I found that if the generated video's time exceeds 15 seconds, the output is always close to the motion frame, and then the error accumulates and gets worse. Referenece frame cannot correct it. During training, does the motion frame need to be noised? do you have any insights?\n\nI would really appreciate it if you could clarify these points. Thank you so much in advance!", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/59/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/58", "id": 3365903116, "node_id": "I_kwDOPdnatc7In5sM", "number": 58, "title": "Lightx2v 4 steps bad result", "user": { "login": "MathewLi9527", "id": 3665731, "node_id": "MDQ6VXNlcjM2NjU3MzE=", "avatar_url": "https://avatars.githubusercontent.com/u/3665731?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MathewLi9527", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 4, "created_at": "2025-08-29T08:08:01Z", "updated_at": "2025-11-18T01:32:43Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "When I use Lightx2v lora with 4 steps, the result is not good. \n\nhttps://github.com/user-attachments/assets/6d8096cd-44f0-47fb-b234-1a3a2493d0cb", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/58/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/57", "id": 3362086758, "node_id": "I_kwDOPdnatc7IZV9m", "number": 57, "title": "Are there any ways to Speed Up Inference?", "user": { "login": "campar", "id": 23508762, "node_id": "MDQ6VXNlcjIzNTA4NzYy", "avatar_url": "https://avatars.githubusercontent.com/u/23508762?v=4", "gravatar_id": "", "url": "https://api.github.com/users/campar", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 4, "created_at": "2025-08-28T07:02:14Z", "updated_at": "2025-09-02T20:16:14Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Is it possible to speed up time inference in any way. I've tried FusionX LORA, but it's almost the same. Is it possible to run inference on multiple GPUs in parallel?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/57/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/56", "id": 3361732866, "node_id": "I_kwDOPdnatc7IX_kC", "number": 56, "title": "Process killed mid-way error when using comfyUI running infiniteTalk", "user": { "login": "NovelX-KamenRider", "id": 220437295, "node_id": "U_kgDODSObLw", "avatar_url": "https://avatars.githubusercontent.com/u/220437295?v=4", "gravatar_id": "", "url": "https://api.github.com/users/NovelX-KamenRider", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-28T04:22:05Z", "updated_at": "2025-09-01T07:51:29Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "when i run infiniteTalk to create a video with audio about 1 minute 30 seconds long, every time it reaches frame size 1000 the process will be killed even though i reduce the image size to 512x512 using gpu rtx4090 and also increase numframe in node InfiniteIalkwave2vecEmbeds to 2500 frames\n\n\"Image\"", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/56/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/55", "id": 3361391761, "node_id": "I_kwDOPdnatc7IWsSR", "number": 55, "title": "不支持中文吗？", "user": { "login": "xtbwqtq", "id": 146714110, "node_id": "U_kgDOCL6t_g", "avatar_url": "https://avatars.githubusercontent.com/u/146714110?v=4", "gravatar_id": "", "url": "https://api.github.com/users/xtbwqtq", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-28T01:17:59Z", "updated_at": "2025-08-28T02:10:43Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "是否支持中文呢？", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/55/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/52", "id": 3358445051, "node_id": "I_kwDOPdnatc7ILc37", "number": 52, "title": "Redundant code when inference", "user": { "login": "1311894932", "id": 126459703, "node_id": "U_kgDOB4mfNw", "avatar_url": "https://avatars.githubusercontent.com/u/126459703?v=4", "gravatar_id": "", "url": "https://api.github.com/users/1311894932", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-27T08:04:57Z", "updated_at": "2025-09-04T08:39:42Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "In the inference code: \nhttps://github.com/MeiGen-AI/InfiniteTalk/blob/0107cd96421b3b1bef7728bde14447dbff7b2db2/wan/multitalk.py#L694\n\nBut after adding noise on the prefix of latent, the prefix will be replaced:\nhttps://github.com/MeiGen-AI/InfiniteTalk/blob/0107cd96421b3b1bef7728bde14447dbff7b2db2/wan/multitalk.py#L711\n\nThus, adding noise is meaningless, the same problem is at:\nhttps://github.com/MeiGen-AI/InfiniteTalk/blob/0107cd96421b3b1bef7728bde14447dbff7b2db2/wan/multitalk.py#L766\nhttps://github.com/MeiGen-AI/InfiniteTalk/blob/0107cd96421b3b1bef7728bde14447dbff7b2db2/wan/multitalk.py#L773\n\nAny response will be appreciated !!", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/52/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/49", "id": 3354685896, "node_id": "I_kwDOPdnatc7H9HHI", "number": 49, "title": "请问可以使用gguf版本的wan吗", "user": { "login": "Aliang-CN", "id": 35071040, "node_id": "MDQ6VXNlcjM1MDcxMDQw", "avatar_url": "https://avatars.githubusercontent.com/u/35071040?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Aliang-CN", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-26T08:29:44Z", "updated_at": "2025-09-02T05:07:46Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "请问可以使用gguf版本的wan吗", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/49/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/48", "id": 3354453782, "node_id": "I_kwDOPdnatc7H8OcW", "number": 48, "title": "Bug caused wrong lip shape in the end of the frames?", "user": { "login": "supermeng", "id": 3438462, "node_id": "MDQ6VXNlcjM0Mzg0NjI=", "avatar_url": "https://avatars.githubusercontent.com/u/3438462?v=4", "gravatar_id": "", "url": "https://api.github.com/users/supermeng", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-08-26T07:17:31Z", "updated_at": "2025-08-26T07:17:31Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "https://github.com/MeiGen-AI/InfiniteTalk/blob/0107cd96421b3b1bef7728bde14447dbff7b2db2/wan/multitalk.py#L844\n\nall of the comfyui workflow would leave the wrong lip shape in the end of the frames, It's the defect of InfiniteTalk or the mis-restoration of comfyui https://github.com/kijai/ComfyUI-WanVideoWrapper?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/48/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/46", "id": 3354070509, "node_id": "I_kwDOPdnatc7H6w3t", "number": 46, "title": "Is there any discussion group about infinite talk or multi talk?", "user": { "login": "chpyang0229", "id": 21274001, "node_id": "MDQ6VXNlcjIxMjc0MDAx", "avatar_url": "https://avatars.githubusercontent.com/u/21274001?v=4", "gravatar_id": "", "url": "https://api.github.com/users/chpyang0229", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-08-26T04:46:18Z", "updated_at": "2025-08-26T04:52:55Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Thanks the author's contribution. \nI encountered some issues, during the implementation process and was wondering if there is a group I can join.\nI am not sure the authors are willing to create a discussion group, \nif not, I’ve created one — everyone interested is welcome to join! \n\n\n\n![Image](https://github.com/user-attachments/assets/bfaac7df-c32f-47c6-b103-12c9862b2efd)", "closed_by": { "login": "chpyang0229", "id": 21274001, "node_id": "MDQ6VXNlcjIxMjc0MDAx", "avatar_url": "https://avatars.githubusercontent.com/u/21274001?v=4", "gravatar_id": "", "url": "https://api.github.com/users/chpyang0229", "type": "User", "user_view_type": "public", "site_admin": false }, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/46/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": "reopened", "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/45", "id": 3354002540, "node_id": "I_kwDOPdnatc7H6gRs", "number": 45, "title": "Sage Attention", "user": { "login": "Redtash1", "id": 175253144, "node_id": "U_kgDOCnImmA", "avatar_url": "https://avatars.githubusercontent.com/u/175253144?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Redtash1", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-08-26T04:06:52Z", "updated_at": "2025-08-26T07:51:49Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "I use Windows 11, Python 3.10.9, Torch 2.7.1+cu128, 4060Ti 16VRAM & 64GB Ram. Would it be possible for you to add Sage Attention 2++ support as there are pre-built wheels that work, \"which i have working in Wan2GP\" but the pre-built Flash Attention wheels on Huggingface no longer work because of a naming error? Thank you.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/45/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/43", "id": 3353165522, "node_id": "I_kwDOPdnatc7H3T7S", "number": 43, "title": "Can we use Quantized Fusion X Lora for multitalk with the Quant InfiniteTalk model?", "user": { "login": "wlsdml1114", "id": 6532977, "node_id": "MDQ6VXNlcjY1MzI5Nzc=", "avatar_url": "https://avatars.githubusercontent.com/u/6532977?v=4", "gravatar_id": "", "url": "https://api.github.com/users/wlsdml1114", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-25T20:45:04Z", "updated_at": "2025-08-27T01:47:31Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "As i see there is no code for quant lora for quant model \n\n```\n# line 187, /InfiniteTalk/wan/multitalk.py\n if quant is not None:\n logging.info(f\"Loading Quantized MultiTalk from {quant_dir}\")\n with torch.device('meta'):\n wan_config = json.load(open(os.path.join(checkpoint_dir, \"config.json\")))\n self.model = WanModel(weight_init=False,**wan_config)\n torch_gc()\n model_state_dict = load_file(quant_dir)\n map_json_path = os.path.join(quant_dir.replace('safetensors', 'json'))\n self.model.init_freqs()\n with open(map_json_path, \"r\") as f:\n quantization_map = json.load(f)\n requantize(self.model, model_state_dict, quantization_map, device='cpu')\n```", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/43/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/42", "id": 3352991098, "node_id": "I_kwDOPdnatc7H2pV6", "number": 42, "title": "ComfyUI Native nodes for 8vram", "user": { "login": "polymath-wtf", "id": 147452692, "node_id": "U_kgDOCMnzFA", "avatar_url": "https://avatars.githubusercontent.com/u/147452692?v=4", "gravatar_id": "", "url": "https://api.github.com/users/polymath-wtf", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-08-25T19:37:28Z", "updated_at": "2025-08-26T20:12:57Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hi, could you please make nodes for native workflow in ComfyUI?\nKijai doing good job with her wrapper nodes, but he are usable only with high VRAM. \nI have 3070ti 8vram and 64ram, and even img2vid wrapper dont work stable 480x480 25f, cause of gguf optimization\nWhen on native flow i can run 1280x720 45frames or 1024x576 121 frame for ez, with [MultiGpu](https://github.com/pollockjj/ComfyUI-MultiGPU.git) lifehack with donor -24ram => 32VRAM\nCould you please make nodes for native use? Cause i think like 90% users have 8vram or lower, and cant do anything with cool stuff like that\nIt seems like we need audio encoder node and new hole for audio embed in WanImageToVideo node \nIt sounds like perfect pleasure for next update by @comfyanonymous 🤗😅", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/42/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/41", "id": 3350800631, "node_id": "I_kwDOPdnatc7HuSj3", "number": 41, "title": "huggingface-cli download MeiGen-AI/InfiniteTalk --local-dir ./weights/InfiniteTalk ---- failed", "user": { "login": "sitexa", "id": 7471477, "node_id": "MDQ6VXNlcjc0NzE0Nzc=", "avatar_url": "https://avatars.githubusercontent.com/u/7471477?v=4", "gravatar_id": "", "url": "https://api.github.com/users/sitexa", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-25T07:41:39Z", "updated_at": "2025-08-26T01:57:13Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "huggingface-cli download MeiGen-AI/InfiniteTalk --local-dir ./weights/InfiniteTalk，下载进度在99%处卡住了几天没有动，如何解决？", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/41/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/40", "id": 3350789382, "node_id": "I_kwDOPdnatc7HuP0G", "number": 40, "title": "Question about calculate_x_ref_attn_map", "user": { "login": "GallenShao", "id": 12490086, "node_id": "MDQ6VXNlcjEyNDkwMDg2", "avatar_url": "https://avatars.githubusercontent.com/u/12490086?v=4", "gravatar_id": "", "url": "https://api.github.com/users/GallenShao", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-08-25T07:37:08Z", "updated_at": "2025-08-25T07:37:08Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Thank you for your great work—the mask-extraction method in self-attn is very insightful and inspiring.\n\nIn [line](https://github.com/MeiGen-AI/InfiniteTalk/blob/0107cd96421b3b1bef7728bde14447dbff7b2db2/wan/utils/multitalk_utils.py#L103), `x_ref_attnmapis` divided by `ref_target_mask.sum()`, which significantly reduces its magnitude and may lead to loss of numerical precision, especially when using torch.float16 or even lower. However, in [line](https://github.com/MeiGen-AI/InfiniteTalk/blob/0107cd96421b3b1bef7728bde14447dbff7b2db2/wan/modules/attention.py#L347), `x_ref_attn_map` is scaled back. Could you clarify the reasoning behind this design choice? Is there a specific consideration for implementing this two-step scaling process?\n\nIn fact, the values of `x_ref_attnmap.sum(-1)` already fall within the range [0, 1], which appears to adequately represent the relevance of each token to the reference frame mask. This further raises the question about the necessity of the additional scaling operation.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/40/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/39", "id": 3350437603, "node_id": "I_kwDOPdnatc7Hs57j", "number": 39, "title": "一运行就报killed什么原因？", "user": { "login": "converxu", "id": 20041815, "node_id": "MDQ6VXNlcjIwMDQxODE1", "avatar_url": "https://avatars.githubusercontent.com/u/20041815?v=4", "gravatar_id": "", "url": "https://api.github.com/users/converxu", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 4, "created_at": "2025-08-25T05:14:47Z", "updated_at": "2025-08-26T08:28:23Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "(multitalk) root@autodl-container-f42141b692-393b4407:~/autodl-tmp/InfiniteTalk# python generate_infinitetalk.py --ckpt_dir weights/Wan2.1-I2V-14B-480P --wav2vec_dir 'weights/chinese-wav2vec2-base' --infinitetalk_dir weights/InfiniteTalk/single/infinitetalk.safetensors --input_json examples/single_example_image.json --size infinitetalk-480 --sample_steps 40 --mode streaming --motion_frame 9 --save_file infinitetalk_res\n[2025-08-25 13:04:40,831] INFO: offload_model is not specified, set to True.\n[2025-08-25 13:04:40,831] INFO: Generation job args: Namespace(task='infinitetalk-14B', size='infinitetalk-480', frame_num=81, max_frame_num=1000, ckpt_dir='weights/Wan2.1-I2V-14B-480P', infinitetalk_dir='weights/InfiniteTalk/single/infinitetalk.safetensors', quant_dir=None, wav2vec_dir='weights/chinese-wav2vec2-base', dit_path=None, lora_dir=None, lora_scale=[1.2], offload_model=True, ulysses_size=1, ring_size=1, t5_fsdp=False, t5_cpu=False, dit_fsdp=False, save_file='infinitetalk_res', audio_save_dir='save_audio', base_seed=42, input_json='examples/single_example_image.json', motion_frame=9, mode='streaming', sample_steps=40, sample_shift=7, sample_text_guide_scale=5.0, sample_audio_guide_scale=4.0, num_persistent_param_in_dit=None, audio_mode='localfile', use_teacache=False, teacache_thresh=0.2, use_apg=False, apg_momentum=-0.75, apg_norm_threshold=55, color_correction_strength=1.0, scene_seg=False, quant=None)\n[2025-08-25 13:04:40,831] INFO: Generation model config: {'__name__': 'Config: Wan MultiTalk AI2V 14B', 't5_model': 'umt5_xxl', 't5_dtype': torch.bfloat16, 'text_len': 512, 'param_dtype': torch.bfloat16, 'num_train_timesteps': 1000, 'sample_fps': 16, 'sample_neg_prompt': 'bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards', 't5_checkpoint': 'models_t5_umt5-xxl-enc-bf16.pth', 't5_tokenizer': 'google/umt5-xxl', 'clip_model': 'clip_xlm_roberta_vit_h_14', 'clip_dtype': torch.float16, 'clip_checkpoint': 'models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth', 'clip_tokenizer': 'xlm-roberta-large', 'vae_checkpoint': 'Wan2.1_VAE.pth', 'vae_stride': (4, 8, 8), 'patch_size': (1, 2, 2), 'dim': 5120, 'ffn_dim': 13824, 'freq_dim': 256, 'num_heads': 40, 'num_layers': 40, 'window_size': (-1, -1), 'qk_norm': True, 'cross_attn_norm': True, 'eps': 1e-06}\n[2025-08-25 13:04:40,831] INFO: Creating infinitetalk pipeline.\n[2025-08-25 13:04:40,831] INFO: loading weights/Wan2.1-I2V-14B-480P/models_t5_umt5-xxl-enc-bf16.pth\n[2025-08-25 13:05:57,211] INFO: loading weights/Wan2.1-I2V-14B-480P/Wan2.1_VAE.pth\n[2025-08-25 13:05:57,791] INFO: loading weights/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth\n[2025-08-25 13:06:02,697] INFO: Creating WanModel from weights/Wan2.1-I2V-14B-480P\nKilled", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/39/reactions", "total_count": 1, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 1, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/38", "id": 3350350111, "node_id": "I_kwDOPdnatc7Hskkf", "number": 38, "title": "Use a video as input, the output is almost static", "user": { "login": "wangfei1204", "id": 15335622, "node_id": "MDQ6VXNlcjE1MzM1NjIy", "avatar_url": "https://avatars.githubusercontent.com/u/15335622?v=4", "gravatar_id": "", "url": "https://api.github.com/users/wangfei1204", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-25T04:17:43Z", "updated_at": "2025-08-26T07:34:12Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "First, thanks a lot for open sourcing this wonderful work. \n\nI use an video(silent video) as input, expect the output video following the input video's motion, and synchronize the lip. But the result video is almost static, only the head has some motion. Does only the first frame of the input video effect the result?\n\nThanks again.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/38/reactions", "total_count": 3, "+1": 3, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/37", "id": 3350035725, "node_id": "I_kwDOPdnatc7HrX0N", "number": 37, "title": "請問支援50系列顯示卡嗎?", "user": { "login": "YOYOKURIHUANG", "id": 202882331, "node_id": "U_kgDODBe9Gw", "avatar_url": "https://avatars.githubusercontent.com/u/202882331?v=4", "gravatar_id": "", "url": "https://api.github.com/users/YOYOKURIHUANG", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-08-25T00:16:15Z", "updated_at": "2025-08-29T11:57:44Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "請問支援50系列顯示卡嗎? 謝謝您", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/37/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/36", "id": 3349993674, "node_id": "I_kwDOPdnatc7HrNjK", "number": 36, "title": "Infinitetalk support in Wan2GP, fast and only needs 8 GB of VRAM to generate 81 frames at time", "user": { "login": "deepbeepmeep", "id": 84379123, "node_id": "MDQ6VXNlcjg0Mzc5MTIz", "avatar_url": "https://avatars.githubusercontent.com/u/84379123?v=4", "gravatar_id": "", "url": "https://api.github.com/users/deepbeepmeep", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 4, "created_at": "2025-08-24T23:16:11Z", "updated_at": "2025-08-27T00:28:26Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Many thanks for this project which is my favorite extension for Wan 2.1 with Vace.\n\nSupport has been added in WanGP (https://github.com/deepbeepmeep/Wan2GP/). WanGP is optimized for low VRAM and offers many video edtiting option and other models (MMaudio support, Qwen Image Edit, ...)\n@yzhang2016 , please let your users know about. I am sure lot will appreciate to be able to use Infinitetalk on low end systems. ", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/36/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/33", "id": 3349365661, "node_id": "I_kwDOPdnatc7Ho0Od", "number": 33, "title": "DownloadAndLoadWav2VecModel", "user": { "login": "raochinmay6", "id": 160831464, "node_id": "U_kgDOCZYX6A", "avatar_url": "https://avatars.githubusercontent.com/u/160831464?v=4", "gravatar_id": "", "url": "https://api.github.com/users/raochinmay6", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-24T09:58:05Z", "updated_at": "2025-08-24T09:58:34Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Prompt executed in 0.08 seconds\nFETCH ComfyRegistry Data: 85/95\nFETCH ComfyRegistry Data: 90/95\nFETCH ComfyRegistry Data: 95/95\nFETCH ComfyRegistry Data [DONE]\n[ComfyUI-Manager] default cache updated: https://api.comfy.org/nodes\nFETCH DATA from: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json [DONE]\n[ComfyUI-Manager] All startup tasks have been completed.\ngot prompt\nFailed to validate prompt for output 131:\n* (prompt):\n - Value not in list: format: 'video/nvenc_h264-mp4-old' not in ['image/gif', 'image/webp', 'video/16bit-png', 'video/8bit-png', 'video/av1-webm', 'video/ffmpeg-gif', 'video/ffv1-mkv', 'video/h264-mp4', 'video/h265-mp4', 'video/nvenc_av1-mp4', 'video/nvenc_h264-mp4', 'video/nvenc_hevc-mp4', 'video/ProRes', 'video/webm']\n* VHS_VideoCombine 131:\n - Value not in list: format: 'video/nvenc_h264-mp4-old' not in ['image/gif', 'image/webp', 'video/16bit-png', 'video/8bit-png', 'video/av1-webm', 'video/ffmpeg-gif', 'video/ffv1-mkv', 'video/h264-mp4', 'video/h265-mp4', 'video/nvenc_av1-mp4', 'video/nvenc_h264-mp4', 'video/nvenc_hevc-mp4', 'video/ProRes', 'video/webm']\nOutput will be ignored\n!!! Exception during processing !!! Due to a serious vulnerability issue in `torch.load`, even with `weights_only=True`, we now require users to upgrade torch to at least v2.6 in order to use the function. This version restriction does not apply when loading files with safetensors.\nSee the vulnerability report here https://nvd.nist.gov/vuln/detail/CVE-2025-32434\nTraceback (most recent call last):\n File \"D:\\ComfyUI_windows_portable\\ComfyUI\\execution.py\", line 496, in execute\n output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, hidden_inputs=hidden_inputs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"D:\\ComfyUI_windows_portable\\ComfyUI\\execution.py\", line 315, in get_output_data\n return_values = await _async_map_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, hidden_inputs=hidden_inputs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"D:\\ComfyUI_windows_portable\\ComfyUI\\execution.py\", line 289, in _async_map_node_over_list\n await process_inputs(input_dict, i)\n File \"D:\\ComfyUI_windows_portable\\ComfyUI\\execution.py\", line 277, in process_inputs\n result = f(**inputs)\n ^^^^^^^^^^^\n File \"D:\\ComfyUI_windows_portable\\ComfyUI\\custom_nodes\\ComfyUI-WanVideoWrapper\\fantasytalking\\nodes.py\", line 72, in loadmodel\n wav2vec = MultiTalkWav2Vec2Model.from_pretrained(model_path).to(base_dtype).to(transfomer_load_device).eval()\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"D:\\ComfyUI_windows_portable\\python_embeded\\Lib\\site-packages\\transformers\\modeling_utils.py\", line 311, in _wrapper\n return func(*args, **kwargs)\n ^^^^^^^^^^^^^^^^^^^^^\n File \"D:\\ComfyUI_windows_portable\\python_embeded\\Lib\\site-packages\\transformers\\modeling_utils.py\", line 4833, in from_pretrained\n ) = cls._load_pretrained_model(\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"D:\\ComfyUI_windows_portable\\python_embeded\\Lib\\site-packages\\transformers\\modeling_utils.py\", line 5099, in _load_pretrained_model\n load_state_dict(checkpoint_files[0], map_location=\"meta\", weights_only=weights_only).keys()\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"D:\\ComfyUI_windows_portable\\python_embeded\\Lib\\site-packages\\transformers\\modeling_utils.py\", line 556, in load_state_dict\n check_torch_load_is_safe()\n File \"D:\\ComfyUI_windows_portable\\python_embeded\\Lib\\site-packages\\transformers\\utils\\import_utils.py\", line 1517, in check_torch_load_is_safe\n raise ValueError(\nValueError: Due to a serious vulnerability issue in `torch.load`, even with `weights_only=True`, we now require users to upgrade torch to at least v2.6 in order to use the function. This version restriction does not apply when loading files with safetensors.\nSee the vulnerability report here https://nvd.nist.gov/vuln/detail/CVE-2025-32434\n\nPrompt executed in 0.05 seconds\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/33/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/32", "id": 3348300485, "node_id": "I_kwDOPdnatc7HkwLF", "number": 32, "title": "xFormers 0.0.28 not found", "user": { "login": "HiTech24SW", "id": 163492511, "node_id": "U_kgDOCb6ynw", "avatar_url": "https://avatars.githubusercontent.com/u/163492511?v=4", "gravatar_id": "", "url": "https://api.github.com/users/HiTech24SW", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-08-23T14:21:00Z", "updated_at": "2025-08-29T02:49:22Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "\n(multitalk) C:\\InfiniteTalk>pip install -U xformers==0.0.28 --index-url https://download.pytorch.org/whl/cu121\nLooking in indexes: https://download.pytorch.org/whl/cu121\n\nERROR: Could not find a version that satisfies the requirement xformers==0.0.28 (from versions: 0.0.22.post4, 0.0.22.post7, 0.0.23, 0.0.23.post1, 0.0.24, 0.0.25, 0.0.25.post1, 0.0.26, 0.0.26.post1, 0.0.27, 0.0.27.post1, 0.0.27.post2)\n\nERROR: No matching distribution found for xformers==0.0.28", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/32/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/31", "id": 3347705096, "node_id": "I_kwDOPdnatc7Hie0I", "number": 31, "title": "Camera Shake Issue.", "user": { "login": "LeCookieGr", "id": 128179391, "node_id": "U_kgDOB6Pcvw", "avatar_url": "https://avatars.githubusercontent.com/u/128179391?v=4", "gravatar_id": "", "url": "https://api.github.com/users/LeCookieGr", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-08-23T07:07:27Z", "updated_at": "2025-09-13T13:29:21Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hello, first of all i want to say that i've been using Multitalk for the last 3 months and im pretty amazed already from it.\nNow this tool seems to be much more cooherent and better BUT.\n\nNo matter how i prompt, the camera always seems to be shaking a bit like if someone does it on purpose.\nI need to know if its possible to have the control over that just like i did with multitalk.\n\nIm currently using Kijai's Workflow.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/31/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/30", "id": 3347073015, "node_id": "I_kwDOPdnatc7HgEf3", "number": 30, "title": "Is the RTX 2080 Ti graphics card (with 11GB VRAM) supported?", "user": { "login": "yhidaoyi", "id": 228124338, "node_id": "U_kgDODZjmsg", "avatar_url": "https://avatars.githubusercontent.com/u/228124338?v=4", "gravatar_id": "", "url": "https://api.github.com/users/yhidaoyi", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-22T23:31:07Z", "updated_at": "2025-08-23T03:37:38Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "I always encounter errors when running ComfyUI. Could you please tell me if it supports the RTX 2080 Ti graphics card (with 11GB VRAM)?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/30/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/28", "id": 3344920794, "node_id": "I_kwDOPdnatc7HX3Da", "number": 28, "title": "Glitchy Last Frame When Using InfiniteTalk", "user": { "login": "Isi-dev", "id": 86603298, "node_id": "MDQ6VXNlcjg2NjAzMjk4", "avatar_url": "https://avatars.githubusercontent.com/u/86603298?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Isi-dev", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-08-22T10:04:14Z", "updated_at": "2025-08-22T10:10:01Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "First, thank you for the great work on InfiniteTalk.\n\nI’ve been testing both MultiTalk and InfiniteTalk in a Colab notebook I put together - https://colab.research.google.com/github/Isi-dev/Google-Colab_Notebooks/blob/main/wan_2_1/wan21_based_InfiniteTalk_&_MultiTalk.ipynb\nWhile running the notebook which is based on Kijai's ComfyUI-WanVideoWrapper, I consistently notice a glitchy last frame at the end of generated videos. This only happens with InfiniteTalk, I have not seen it with Multi-Talk.\nIs this a known/common issue, or could it be related to the models/settings I used?\n\nThanks again for sharing such impressive projects.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/28/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/27", "id": 3344689519, "node_id": "I_kwDOPdnatc7HW-lv", "number": 27, "title": "Test result issues", "user": { "login": "571129857", "id": 31714050, "node_id": "MDQ6VXNlcjMxNzE0MDUw", "avatar_url": "https://avatars.githubusercontent.com/u/31714050?v=4", "gravatar_id": "", "url": "https://api.github.com/users/571129857", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 3, "created_at": "2025-08-22T08:47:33Z", "updated_at": "2025-08-25T01:53:50Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "When I input a video and 20 seconds of audio, the generated video jumps and becomes discontinuous every few seconds. Including the given demo effect, this issue also exists. Is this reasonable? The parameters still haven't been adjusted properly.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/27/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/26", "id": 3344630272, "node_id": "I_kwDOPdnatc7HWwIA", "number": 26, "title": "Chinese audio, lip syncing mismatch", "user": { "login": "Liaooo-liya", "id": 62998228, "node_id": "MDQ6VXNlcjYyOTk4MjI4", "avatar_url": "https://avatars.githubusercontent.com/u/62998228?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Liaooo-liya", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-22T08:26:27Z", "updated_at": "2025-08-23T01:01:00Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "I used the image-to-video method, using KJ's workflow, to generate a 5-second video with Chinese audio. However, the results were less than ideal. The lip syncing started to mismatch in the second half of the video.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/26/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/25", "id": 3344374087, "node_id": "I_kwDOPdnatc7HVxlH", "number": 25, "title": "Does this work with 12GB VRAM?", "user": { "login": "24hourdiscordbot", "id": 104885713, "node_id": "U_kgDOBkBt0Q", "avatar_url": "https://avatars.githubusercontent.com/u/104885713?v=4", "gravatar_id": "", "url": "https://api.github.com/users/24hourdiscordbot", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-22T06:56:21Z", "updated_at": "2025-08-24T23:11:08Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "I noticed in the README that low VRAM support is mentioned. I'm wondering if this would work smoothly with my 12GB VRAM setup?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/25/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/24", "id": 3344320623, "node_id": "I_kwDOPdnatc7HVkhv", "number": 24, "title": "Training LoRA/fine tuning for other languages", "user": { "login": "binlebin", "id": 928869, "node_id": "MDQ6VXNlcjkyODg2OQ==", "avatar_url": "https://avatars.githubusercontent.com/u/928869?v=4", "gravatar_id": "", "url": "https://api.github.com/users/binlebin", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 5, "created_at": "2025-08-22T06:32:57Z", "updated_at": "2025-09-08T20:47:08Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "First, thanks a lot for open sourcing this work. This is phenomenal! I have followed the project since MultiTalk.\n\nI just tested the model with Vietnamese audio and the lipsync quality is average. I think additional training required. \n\nCan you help point me to the right way to improve this? Training additional LoRA or fine tuning the model further? Do you plan to share any training code? (Or perhaps point me to the right path to learn?).\n\nThanks again!\n\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/24/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/22", "id": 3343024845, "node_id": "I_kwDOPdnatc7HQoLN", "number": 22, "title": "the install instructions for ComfyUI usage are either incorrect or unclear", "user": { "login": "bsenftner", "id": 1216815, "node_id": "MDQ6VXNlcjEyMTY4MTU=", "avatar_url": "https://avatars.githubusercontent.com/u/1216815?v=4", "gravatar_id": "", "url": "https://api.github.com/users/bsenftner", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-08-21T20:00:11Z", "updated_at": "2025-08-21T20:02:17Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Then viewing the ComfyUI branch of this repo\n\nhttps://github.com/MeiGen-AI/InfiniteTalk/tree/comfyui\n\nThe ComfyUI wrapper nodes for [InfiniteTalk](https://github.com/MeiGen-AI/InfiniteTalk) installation instructions simply say for step 1:\n\n1> Clone this repo into custom_nodes folder.\n\nThis is not enough information, because blindly cloning the repo gives one the main branch, which is incompatible with ComfyUI.\n\nOne needs to either switch to the \"comfyui\" branch after cloning, or one of:\n\n`git clone --single-branch --branch `\n \nor\n\n`git clone -b `\n\nEssentially, the ComfyUI installation needs to instruct to only clone the \"comfyui\" branch or to explain better and instruct to switch to the \"comfyui\" branch after cloning and before launching the ComfyUI application. ", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/22/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/21", "id": 3342978499, "node_id": "I_kwDOPdnatc7HQc3D", "number": 21, "title": "Is infiniteTalk support sage attention2.2?", "user": { "login": "wlsdml1114", "id": 6532977, "node_id": "MDQ6VXNlcjY1MzI5Nzc=", "avatar_url": "https://avatars.githubusercontent.com/u/6532977?v=4", "gravatar_id": "", "url": "https://api.github.com/users/wlsdml1114", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-08-21T19:44:52Z", "updated_at": "2025-08-21T19:44:52Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Thank you for great job!\n\nMultitalk supports Sage Attention 2.2. Is there a plan to support Sage Attention 2.2 with InfiniteTalk? Or does it already support it?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/21/reactions", "total_count": 1, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 1 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/19", "id": 3340923183, "node_id": "I_kwDOPdnatc7HInEv", "number": 19, "title": "Questions on context frames and reference frame in paper vs. code implementation", "user": { "login": "GallenShao", "id": 12490086, "node_id": "MDQ6VXNlcjEyNDkwMDg2", "avatar_url": "https://avatars.githubusercontent.com/u/12490086?v=4", "gravatar_id": "", "url": "https://api.github.com/users/GallenShao", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-21T09:05:36Z", "updated_at": "2025-08-21T12:58:11Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hi, thanks for your great work!\n\nI’d like to ask about the correspondence between the **context frames** and **reference frame** described in the paper, and the actual implementation in the code. Could you please help confirm whether my understanding is correct?\n\n- Context frames correspond to _latent_motion_frames_ in the code. These appear to remain completely fixed during inference.\n- Reference frame corresponds to _cond_image_. In Wan2.1 i2v, the first frame is forced to be exactly aligned with the reference frame, but in your implementation it seems not to be pixel-aligned—instead, only certain features of the image are referenced.\n\nIf this interpretation is correct, I have two further questions:\n\n- The paper doesn’t seem to specify the length of context frames during training and inference. Is this length variable?\n- In Fig. 3 and Fig. 4 of the paper, the context frames do not appear to remain unchanged, which is a bit confusing. Could you clarify this?\n\nThanks a lot!\n\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/19/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/18", "id": 3340764978, "node_id": "I_kwDOPdnatc7HIAcy", "number": 18, "title": "quant model inference speed", "user": { "login": "ljiqy", "id": 37796840, "node_id": "MDQ6VXNlcjM3Nzk2ODQw", "avatar_url": "https://avatars.githubusercontent.com/u/37796840?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ljiqy", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 6, "created_at": "2025-08-21T08:15:52Z", "updated_at": "2025-08-30T06:14:39Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "I tried using FP8 and INT8 models, but the speed actually became slower. On an A800 server, the standard model took 1.3 hours for 40 steps, while the FP8 or INT8 model took 5 hours for 40 steps. Why is this happening?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/18/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/17", "id": 3340732681, "node_id": "I_kwDOPdnatc7HH4kJ", "number": 17, "title": "request to add model on replicate", "user": { "login": "PATCHOULISEN", "id": 89085326, "node_id": "MDQ6VXNlcjg5MDg1MzI2", "avatar_url": "https://avatars.githubusercontent.com/u/89085326?v=4", "gravatar_id": "", "url": "https://api.github.com/users/PATCHOULISEN", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-08-21T08:05:12Z", "updated_at": "2025-08-21T08:05:12Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "let more users can try the model easliy :)", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/17/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/15", "id": 3340156746, "node_id": "I_kwDOPdnatc7HFr9K", "number": 15, "title": "Compare with StableAvatar", "user": { "login": "great-luao", "id": 116716723, "node_id": "U_kgDOBvT0sw", "avatar_url": "https://avatars.githubusercontent.com/u/116716723?v=4", "gravatar_id": "", "url": "https://api.github.com/users/great-luao", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 3, "created_at": "2025-08-21T03:31:27Z", "updated_at": "2025-08-21T07:14:16Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Will there be a comparison between InfiniteTalk and StableAvatar(https://github.com/Francis-Rings/StableAvatar) in the video generation quality?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/15/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/14", "id": 3338709287, "node_id": "I_kwDOPdnatc7HAKkn", "number": 14, "title": "Creating WanModel from weights results in \"Killed\"", "user": { "login": "SuperMaximus1984", "id": 62514634, "node_id": "MDQ6VXNlcjYyNTE0NjM0", "avatar_url": "https://avatars.githubusercontent.com/u/62514634?v=4", "gravatar_id": "", "url": "https://api.github.com/users/SuperMaximus1984", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 6, "created_at": "2025-08-20T16:10:32Z", "updated_at": "2025-08-23T23:20:52Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Regardless of the way it's loaded (Gradio or just a single-GPU inference script) the CLI says \"Killed\" when creates WanModel \n```\n[2025-08-20 19:06:32,687] INFO: Creating infinitetalk pipeline.\n[2025-08-20 19:06:32,688] INFO: loading weights/Wan2.1-I2V-14B-480P/models_t5_umt5-xxl-enc-bf16.pth\n[2025-08-20 19:07:08,315] INFO: loading weights/Wan2.1-I2V-14B-480P/Wan2.1_VAE.pth\n[2025-08-20 19:07:08,650] INFO: loading weights/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth\n[2025-08-20 19:07:10,938] INFO: Creating WanModel from weights/Wan2.1-I2V-14B-480P\nKilled\n```\n\nUsing on 1 x RTX4090 VRAM 24 GB + 128 GB RAM", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/14/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/13", "id": 3337104007, "node_id": "I_kwDOPdnatc7G6CqH", "number": 13, "title": "resolution change of the generated frames", "user": { "login": "MathewLi9527", "id": 3665731, "node_id": "MDQ6VXNlcjM2NjU3MzE=", "avatar_url": "https://avatars.githubusercontent.com/u/3665731?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MathewLi9527", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 8, "created_at": "2025-08-20T08:30:57Z", "updated_at": "2025-08-22T01:31:15Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "\"Image\"\n\nThe resolutions of the starting frame and the end frame are different. The later generated frames have higher resolution than the previous generated frames. Why and how to solve this?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/13/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/12", "id": 3337051453, "node_id": "I_kwDOPdnatc7G5109", "number": 12, "title": "ValueError: The `output_attentions` attribute is not supported when using the `attn_implementation` set to sdpa. Please set it to 'eager' instead.", "user": { "login": "ljiqy", "id": 37796840, "node_id": "MDQ6VXNlcjM3Nzk2ODQw", "avatar_url": "https://avatars.githubusercontent.com/u/37796840?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ljiqy", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 3, "created_at": "2025-08-20T08:12:53Z", "updated_at": "2025-08-22T12:14:44Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "\"Image\"\n请问这个错误怎么解决", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/12/reactions", "total_count": 1, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 1 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/10", "id": 3336341039, "node_id": "I_kwDOPdnatc7G3IYv", "number": 10, "title": "is I2V limited to 40 seconds?", "user": { "login": "chikiuso", "id": 79965, "node_id": "MDQ6VXNlcjc5OTY1", "avatar_url": "https://avatars.githubusercontent.com/u/79965?v=4", "gravatar_id": "", "url": "https://api.github.com/users/chikiuso", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 13, "created_at": "2025-08-20T02:29:36Z", "updated_at": "2025-09-01T02:01:04Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Good morning\n\n It is really an excellent work, fantastic!! May I ask is the image to video maximum duration limited to 40 seconds? And video to video is unlimited as long as the input video length is long enough to avoid its looping? thanks!\n\nBest wishes,\nJack", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/10/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/8", "id": 3335275513, "node_id": "I_kwDOPdnatc7GzEP5", "number": 8, "title": "Gradio Demo", "user": { "login": "motowntek", "id": 177491214, "node_id": "U_kgDOCpRNDg", "avatar_url": "https://avatars.githubusercontent.com/u/177491214?v=4", "gravatar_id": "", "url": "https://api.github.com/users/motowntek", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-19T18:22:13Z", "updated_at": "2025-08-22T03:05:10Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hi. Can you create a Gradio demo on Huggingface that we can duplicate and use? That would be very useful. \nThanks. ", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/8/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/7", "id": 3334689832, "node_id": "I_kwDOPdnatc7Gw1Qo", "number": 7, "title": "Reference benchmarks", "user": { "login": "kasf666", "id": 202331540, "node_id": "U_kgDODA9VlA", "avatar_url": "https://avatars.githubusercontent.com/u/202331540?v=4", "gravatar_id": "", "url": "https://api.github.com/users/kasf666", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 2, "created_at": "2025-08-19T14:51:59Z", "updated_at": "2025-08-20T01:31:01Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hi Guys, again amazing work!\n\nWould be useful to have inference time benchmarks for video reference and image reference in order to understand what runtime to expect.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/7/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/6", "id": 3334685908, "node_id": "I_kwDOPdnatc7Gw0TU", "number": 6, "title": "Optimal video reference length", "user": { "login": "kasf666", "id": 202331540, "node_id": "U_kgDODA9VlA", "avatar_url": "https://avatars.githubusercontent.com/u/202331540?v=4", "gravatar_id": "", "url": "https://api.github.com/users/kasf666", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 5, "created_at": "2025-08-19T14:50:52Z", "updated_at": "2025-11-22T14:42:16Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Is there any specific video length that is optimal as imput and from which time stamp it's diminishing returns? Thanks.", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/6/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/4", "id": 3333246023, "node_id": "I_kwDOPdnatc7GrUxH", "number": 4, "title": "what difference between ComfyUI single model and MeiGen-AI single model?", "user": { "login": "tjohn88", "id": 116616313, "node_id": "U_kgDOBvNseQ", "avatar_url": "https://avatars.githubusercontent.com/u/116616313?v=4", "gravatar_id": "", "url": "https://api.github.com/users/tjohn88", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 0, "created_at": "2025-08-19T07:29:01Z", "updated_at": "2025-08-19T07:29:01Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "What is the difference between ComfyUI infinitetalk.safetensors (2.71) and MeiGen-AI single infinitetalk.safetensors (9.95 GB)?\nIs there a difference in quality?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/4/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/3", "id": 3332883894, "node_id": "I_kwDOPdnatc7Gp8W2", "number": 3, "title": "Getting \"TypeError: WanVideoSampler.process() got an unexpected keyword argument 'infinitetalk_embeds'\"", "user": { "login": "mdkberry", "id": 810955, "node_id": "MDQ6VXNlcjgxMDk1NQ==", "avatar_url": "https://avatars.githubusercontent.com/u/810955?v=4", "gravatar_id": "", "url": "https://api.github.com/users/mdkberry", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 3, "created_at": "2025-08-19T04:56:39Z", "updated_at": "2025-08-22T07:13:51Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Using your workflows from (comfyui branch) example folder I swapped in the infinite talk single model and using i2v but get this error. I didnt reconnect anything, its as per your wf.\n\n\n\"Image\"\n", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/3/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/2", "id": 3332866604, "node_id": "I_kwDOPdnatc7Gp4Is", "number": 2, "title": "is the implementation same as multitalk?", "user": { "login": "Rudra-ai-coder", "id": 177262225, "node_id": "U_kgDOCpDOkQ", "avatar_url": "https://avatars.githubusercontent.com/u/177262225?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Rudra-ai-coder", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 8, "created_at": "2025-08-19T04:45:15Z", "updated_at": "2025-12-16T08:20:47Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": "Hi, is the architecture and implementation same as multi-talk, will this required and extended implementation for usage in ComfyUI?", "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/2/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/1", "id": 3332751208, "node_id": "I_kwDOPdnatc7Gpb9o", "number": 1, "title": "Hello, the project link and the paper link is not avaliable", "user": { "login": "StellarCheng", "id": 122878624, "node_id": "U_kgDOB1L6oA", "avatar_url": "https://avatars.githubusercontent.com/u/122878624?v=4", "gravatar_id": "", "url": "https://api.github.com/users/StellarCheng", "type": "User", "user_view_type": "public", "site_admin": false }, "labels": {}, "state": "open", "locked": false, "assignee": null, "assignees": {}, "milestone": null, "comments": 1, "created_at": "2025-08-19T03:24:48Z", "updated_at": "2025-08-19T05:15:29Z", "closed_at": null, "author_association": "NONE", "type": null, "active_lock_reason": null, "sub_issues_summary": { "total": 0, "completed": 0, "percent_completed": 0 }, "issue_dependencies_summary": { "blocked_by": 0, "total_blocked_by": 0, "blocking": 0, "total_blocking": 0 }, "body": null, "closed_by": null, "reactions": { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/1/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "performed_via_github_app": null, "state_reason": null, "linked_prs": [ 180 ] } ], "pulls": [ { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/180", "id": 3145858292, "node_id": "PR_kwDOPdnatc67gfz0", "number": 180, "state": "open", "locked": false, "title": "[fix]: resolve performance issue in multi-GPU inference", "user": { "login": "jokerz0624", "id": 50729849, "node_id": "MDQ6VXNlcjUwNzI5ODQ5", "avatar_url": "https://avatars.githubusercontent.com/u/50729849?v=4", "gravatar_id": "", "url": "https://api.github.com/users/jokerz0624", "type": "User", "user_view_type": "public", "site_admin": false }, "body": "#147 \r\noriginal 'split_token_counts_and_frame_ids' function takes long time to run", "created_at": "2026-01-05T10:36:05Z", "updated_at": "2026-01-05T10:36:05Z", "closed_at": null, "merged_at": null, "merge_commit_sha": "d38fd55b93275bd1fea5113180445a880d531b8f", "assignee": null, "assignees": {}, "requested_reviewers": {}, "requested_teams": {}, "labels": {}, "milestone": null, "draft": false, "head": { "label": "jokerz0624:multy-infer", "ref": "multy-infer", "sha": "87adf1cc01ec53854a872a1b62a5fde42ef76f00", "user": { "login": "jokerz0624", "id": 50729849, "node_id": "MDQ6VXNlcjUwNzI5ODQ5", "avatar_url": "https://avatars.githubusercontent.com/u/50729849?v=4", "gravatar_id": "", "url": "https://api.github.com/users/jokerz0624", "type": "User", "user_view_type": "public", "site_admin": false }, "repo": { "id": 1128226772, "node_id": "R_kgDOQz9f1A", "name": "InfiniteTalk", "full_name": "jokerz0624/InfiniteTalk", "private": false, "owner": { "login": "jokerz0624", "id": 50729849, "node_id": "MDQ6VXNlcjUwNzI5ODQ5", "avatar_url": "https://avatars.githubusercontent.com/u/50729849?v=4", "gravatar_id": "", "url": "https://api.github.com/users/jokerz0624", "type": "User", "user_view_type": "public", "site_admin": false }, "description": "​​Unlimited-length talking video generation​​ that supports image-to-video and video-to-video generation", "fork": true, "url": "https://api.github.com/repos/jokerz0624/InfiniteTalk", "created_at": "2026-01-05T10:23:15Z", "updated_at": "2026-01-05T10:29:56Z", "pushed_at": "2026-01-05T10:32:31Z", "homepage": null, "size": 21464, "stargazers_count": 0, "watchers_count": 0, "language": "Python", "has_issues": false, "has_projects": true, "has_downloads": true, "has_wiki": false, "has_pages": false, "has_discussions": false, "forks_count": 0, "archived": false, "disabled": false, "open_issues_count": 0, "license": { "key": "apache-2.0", "name": "Apache License 2.0", "spdx_id": "Apache-2.0", "url": "https://api.github.com/licenses/apache-2.0", "node_id": "MDc6TGljZW5zZTI=" }, "allow_forking": true, "is_template": false, "web_commit_signoff_required": false, "topics": {}, "visibility": "public", "forks": 0, "open_issues": 0, "watchers": 0, "default_branch": "main" } }, "base": { "label": "MeiGen-AI:main", "ref": "main", "sha": "fd631497254e065777f2b2d0642de3600d674e24", "user": { "login": "MeiGen-AI", "id": 213629870, "node_id": "O_kgDODLu7rg", "avatar_url": "https://avatars.githubusercontent.com/u/213629870?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MeiGen-AI", "type": "Organization", "user_view_type": "public", "site_admin": false }, "repo": { "id": 1037687477, "node_id": "R_kgDOPdnatQ", "name": "InfiniteTalk", "full_name": "MeiGen-AI/InfiniteTalk", "private": false, "owner": { "login": "MeiGen-AI", "id": 213629870, "node_id": "O_kgDODLu7rg", "avatar_url": "https://avatars.githubusercontent.com/u/213629870?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MeiGen-AI", "type": "Organization", "user_view_type": "public", "site_admin": false }, "description": "​​Unlimited-length talking video generation​​ that supports image-to-video and video-to-video generation", "fork": false, "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk", "created_at": "2025-08-14T01:04:53Z", "updated_at": "2026-01-13T02:31:17Z", "pushed_at": "2025-12-18T03:48:58Z", "homepage": null, "size": 247145, "stargazers_count": 4376, "watchers_count": 4376, "language": "Python", "has_issues": true, "has_projects": true, "has_downloads": true, "has_wiki": false, "has_pages": true, "has_discussions": false, "forks_count": 731, "archived": false, "disabled": false, "open_issues_count": 148, "license": { "key": "apache-2.0", "name": "Apache License 2.0", "spdx_id": "Apache-2.0", "url": "https://api.github.com/licenses/apache-2.0", "node_id": "MDc6TGljZW5zZTI=" }, "allow_forking": true, "is_template": false, "web_commit_signoff_required": false, "topics": {}, "visibility": "public", "forks": 731, "open_issues": 148, "watchers": 4376, "default_branch": "main" } }, "_links": { "self": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/180" }, "html": { "href": "https://github.com/MeiGen-AI/InfiniteTalk/pull/180" }, "issue": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/180" }, "comments": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/180/comments" }, "review_comments": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/180/comments" }, "review_comment": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/comments{/number}" }, "commits": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/180/commits" }, "statuses": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/statuses/87adf1cc01ec53854a872a1b62a5fde42ef76f00" } }, "author_association": "NONE", "auto_merge": null, "active_lock_reason": null, "linked_issues": [ 1 ] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/178", "id": 3131759374, "node_id": "PR_kwDOPdnatc66qtsO", "number": 178, "state": "open", "locked": false, "title": "Add --no-build-isolation to fix flash_attn build error", "user": { "login": "darouwan", "id": 2324213, "node_id": "MDQ6VXNlcjIzMjQyMTM=", "avatar_url": "https://avatars.githubusercontent.com/u/2324213?v=4", "gravatar_id": "", "url": "https://api.github.com/users/darouwan", "type": "User", "user_view_type": "public", "site_admin": false }, "body": null, "created_at": "2025-12-26T23:42:14Z", "updated_at": "2025-12-26T23:42:14Z", "closed_at": null, "merged_at": null, "merge_commit_sha": "38acbeb3188f6d4002625bb40048aad1b5a9db77", "assignee": null, "assignees": {}, "requested_reviewers": {}, "requested_teams": {}, "labels": {}, "milestone": null, "draft": false, "head": { "label": "darouwan:bugfix/flash_attn_pip", "ref": "bugfix/flash_attn_pip", "sha": "c8de32ed6cea2bf3955280a137926d0a8250f6f0", "user": { "login": "darouwan", "id": 2324213, "node_id": "MDQ6VXNlcjIzMjQyMTM=", "avatar_url": "https://avatars.githubusercontent.com/u/2324213?v=4", "gravatar_id": "", "url": "https://api.github.com/users/darouwan", "type": "User", "user_view_type": "public", "site_admin": false }, "repo": { "id": 1123463561, "node_id": "R_kgDOQvaxiQ", "name": "InfiniteTalk", "full_name": "darouwan/InfiniteTalk", "private": false, "owner": { "login": "darouwan", "id": 2324213, "node_id": "MDQ6VXNlcjIzMjQyMTM=", "avatar_url": "https://avatars.githubusercontent.com/u/2324213?v=4", "gravatar_id": "", "url": "https://api.github.com/users/darouwan", "type": "User", "user_view_type": "public", "site_admin": false }, "description": "​​Unlimited-length talking video generation​​ that supports image-to-video and video-to-video generation", "fork": true, "url": "https://api.github.com/repos/darouwan/InfiniteTalk", "created_at": "2025-12-26T23:36:50Z", "updated_at": "2025-12-26T23:36:50Z", "pushed_at": "2025-12-26T23:40:41Z", "homepage": null, "size": 21475, "stargazers_count": 0, "watchers_count": 0, "language": null, "has_issues": false, "has_projects": true, "has_downloads": true, "has_wiki": false, "has_pages": false, "has_discussions": false, "forks_count": 0, "archived": false, "disabled": false, "open_issues_count": 0, "license": { "key": "apache-2.0", "name": "Apache License 2.0", "spdx_id": "Apache-2.0", "url": "https://api.github.com/licenses/apache-2.0", "node_id": "MDc6TGljZW5zZTI=" }, "allow_forking": true, "is_template": false, "web_commit_signoff_required": false, "topics": {}, "visibility": "public", "forks": 0, "open_issues": 0, "watchers": 0, "default_branch": "main" } }, "base": { "label": "MeiGen-AI:main", "ref": "main", "sha": "fd631497254e065777f2b2d0642de3600d674e24", "user": { "login": "MeiGen-AI", "id": 213629870, "node_id": "O_kgDODLu7rg", "avatar_url": "https://avatars.githubusercontent.com/u/213629870?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MeiGen-AI", "type": "Organization", "user_view_type": "public", "site_admin": false }, "repo": { "id": 1037687477, "node_id": "R_kgDOPdnatQ", "name": "InfiniteTalk", "full_name": "MeiGen-AI/InfiniteTalk", "private": false, "owner": { "login": "MeiGen-AI", "id": 213629870, "node_id": "O_kgDODLu7rg", "avatar_url": "https://avatars.githubusercontent.com/u/213629870?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MeiGen-AI", "type": "Organization", "user_view_type": "public", "site_admin": false }, "description": "​​Unlimited-length talking video generation​​ that supports image-to-video and video-to-video generation", "fork": false, "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk", "created_at": "2025-08-14T01:04:53Z", "updated_at": "2026-01-13T02:31:17Z", "pushed_at": "2025-12-18T03:48:58Z", "homepage": null, "size": 247145, "stargazers_count": 4376, "watchers_count": 4376, "language": "Python", "has_issues": true, "has_projects": true, "has_downloads": true, "has_wiki": false, "has_pages": true, "has_discussions": false, "forks_count": 731, "archived": false, "disabled": false, "open_issues_count": 148, "license": { "key": "apache-2.0", "name": "Apache License 2.0", "spdx_id": "Apache-2.0", "url": "https://api.github.com/licenses/apache-2.0", "node_id": "MDc6TGljZW5zZTI=" }, "allow_forking": true, "is_template": false, "web_commit_signoff_required": false, "topics": {}, "visibility": "public", "forks": 731, "open_issues": 148, "watchers": 4376, "default_branch": "main" } }, "_links": { "self": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/178" }, "html": { "href": "https://github.com/MeiGen-AI/InfiniteTalk/pull/178" }, "issue": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/178" }, "comments": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/178/comments" }, "review_comments": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/178/comments" }, "review_comment": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/comments{/number}" }, "commits": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/178/commits" }, "statuses": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/statuses/c8de32ed6cea2bf3955280a137926d0a8250f6f0" } }, "author_association": "NONE", "auto_merge": null, "active_lock_reason": null, "linked_issues": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/161", "id": 3056043836, "node_id": "PR_kwDOPdnatc62J4c8", "number": 161, "state": "open", "locked": false, "title": "Create run.bat", "user": { "login": "jingtengluo9-hue", "id": 241120745, "node_id": "U_kgDODl816Q", "avatar_url": "https://avatars.githubusercontent.com/u/241120745?v=4", "gravatar_id": "", "url": "https://api.github.com/users/jingtengluo9-hue", "type": "User", "user_view_type": "public", "site_admin": false }, "body": null, "created_at": "2025-11-30T03:03:22Z", "updated_at": "2025-11-30T03:03:22Z", "closed_at": null, "merged_at": null, "merge_commit_sha": "ff822e593ea9cdd874af7cbc1b099926928b24b5", "assignee": null, "assignees": {}, "requested_reviewers": {}, "requested_teams": {}, "labels": {}, "milestone": null, "draft": false, "head": { "label": "jingtengluo9-hue:patch-1", "ref": "patch-1", "sha": "c9c191baa8d33a05fa171d44991b273da7726226", "user": { "login": "jingtengluo9-hue", "id": 241120745, "node_id": "U_kgDODl816Q", "avatar_url": "https://avatars.githubusercontent.com/u/241120745?v=4", "gravatar_id": "", "url": "https://api.github.com/users/jingtengluo9-hue", "type": "User", "user_view_type": "public", "site_admin": false }, "repo": { "id": 1106832173, "node_id": "R_kgDOQfjrLQ", "name": "InfiniteTalk", "full_name": "jingtengluo9-hue/InfiniteTalk", "private": false, "owner": { "login": "jingtengluo9-hue", "id": 241120745, "node_id": "U_kgDODl816Q", "avatar_url": "https://avatars.githubusercontent.com/u/241120745?v=4", "gravatar_id": "", "url": "https://api.github.com/users/jingtengluo9-hue", "type": "User", "user_view_type": "public", "site_admin": false }, "description": "​​Unlimited-length talking video generation​​ that supports image-to-video and video-to-video generation", "fork": true, "url": "https://api.github.com/repos/jingtengluo9-hue/InfiniteTalk", "created_at": "2025-11-30T03:02:00Z", "updated_at": "2025-11-30T03:02:00Z", "pushed_at": "2025-11-30T03:02:39Z", "homepage": null, "size": 247133, "stargazers_count": 0, "watchers_count": 0, "language": null, "has_issues": false, "has_projects": true, "has_downloads": true, "has_wiki": false, "has_pages": false, "has_discussions": false, "forks_count": 0, "archived": false, "disabled": false, "open_issues_count": 0, "license": { "key": "apache-2.0", "name": "Apache License 2.0", "spdx_id": "Apache-2.0", "url": "https://api.github.com/licenses/apache-2.0", "node_id": "MDc6TGljZW5zZTI=" }, "allow_forking": true, "is_template": false, "web_commit_signoff_required": false, "topics": {}, "visibility": "public", "forks": 0, "open_issues": 0, "watchers": 0, "default_branch": "main" } }, "base": { "label": "MeiGen-AI:main", "ref": "main", "sha": "0107cd96421b3b1bef7728bde14447dbff7b2db2", "user": { "login": "MeiGen-AI", "id": 213629870, "node_id": "O_kgDODLu7rg", "avatar_url": "https://avatars.githubusercontent.com/u/213629870?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MeiGen-AI", "type": "Organization", "user_view_type": "public", "site_admin": false }, "repo": { "id": 1037687477, "node_id": "R_kgDOPdnatQ", "name": "InfiniteTalk", "full_name": "MeiGen-AI/InfiniteTalk", "private": false, "owner": { "login": "MeiGen-AI", "id": 213629870, "node_id": "O_kgDODLu7rg", "avatar_url": "https://avatars.githubusercontent.com/u/213629870?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MeiGen-AI", "type": "Organization", "user_view_type": "public", "site_admin": false }, "description": "​​Unlimited-length talking video generation​​ that supports image-to-video and video-to-video generation", "fork": false, "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk", "created_at": "2025-08-14T01:04:53Z", "updated_at": "2026-01-13T02:31:17Z", "pushed_at": "2025-12-18T03:48:58Z", "homepage": null, "size": 247145, "stargazers_count": 4376, "watchers_count": 4376, "language": "Python", "has_issues": true, "has_projects": true, "has_downloads": true, "has_wiki": false, "has_pages": true, "has_discussions": false, "forks_count": 731, "archived": false, "disabled": false, "open_issues_count": 148, "license": { "key": "apache-2.0", "name": "Apache License 2.0", "spdx_id": "Apache-2.0", "url": "https://api.github.com/licenses/apache-2.0", "node_id": "MDc6TGljZW5zZTI=" }, "allow_forking": true, "is_template": false, "web_commit_signoff_required": false, "topics": {}, "visibility": "public", "forks": 731, "open_issues": 148, "watchers": 4376, "default_branch": "main" } }, "_links": { "self": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/161" }, "html": { "href": "https://github.com/MeiGen-AI/InfiniteTalk/pull/161" }, "issue": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/161" }, "comments": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/161/comments" }, "review_comments": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/161/comments" }, "review_comment": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/comments{/number}" }, "commits": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/161/commits" }, "statuses": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/statuses/c9c191baa8d33a05fa171d44991b273da7726226" } }, "author_association": "NONE", "auto_merge": null, "active_lock_reason": null, "linked_issues": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/119", "id": 2862423229, "node_id": "PR_kwDOPdnatc6qnRy9", "number": 119, "state": "open", "locked": false, "title": "Fixed multi-gpu save conflicts", "user": { "login": "CoderMusou", "id": 20904589, "node_id": "MDQ6VXNlcjIwOTA0NTg5", "avatar_url": "https://avatars.githubusercontent.com/u/20904589?v=4", "gravatar_id": "", "url": "https://api.github.com/users/CoderMusou", "type": "User", "user_view_type": "public", "site_admin": false }, "body": "Multiple GPUs will report an error `PytorchStreamReader failed reading zip archive: failed finding central directory`", "created_at": "2025-09-26T03:09:57Z", "updated_at": "2025-11-28T08:14:45Z", "closed_at": null, "merged_at": null, "merge_commit_sha": "9e61d1e4492d6e86563ed55a30e3f3380da99a31", "assignee": null, "assignees": {}, "requested_reviewers": {}, "requested_teams": {}, "labels": {}, "milestone": null, "draft": false, "head": { "label": "CoderMusou:main", "ref": "main", "sha": "15a3d6802229759344af4ce3a94b21bba66d5aa1", "user": { "login": "CoderMusou", "id": 20904589, "node_id": "MDQ6VXNlcjIwOTA0NTg5", "avatar_url": "https://avatars.githubusercontent.com/u/20904589?v=4", "gravatar_id": "", "url": "https://api.github.com/users/CoderMusou", "type": "User", "user_view_type": "public", "site_admin": false }, "repo": { "id": 1064429761, "node_id": "R_kgDOP3HowQ", "name": "InfiniteTalk", "full_name": "CoderMusou/InfiniteTalk", "private": false, "owner": { "login": "CoderMusou", "id": 20904589, "node_id": "MDQ6VXNlcjIwOTA0NTg5", "avatar_url": "https://avatars.githubusercontent.com/u/20904589?v=4", "gravatar_id": "", "url": "https://api.github.com/users/CoderMusou", "type": "User", "user_view_type": "public", "site_admin": false }, "description": "​​Unlimited-length talking video generation​​ that supports image-to-video and video-to-video generation", "fork": true, "url": "https://api.github.com/repos/CoderMusou/InfiniteTalk", "created_at": "2025-09-26T02:59:24Z", "updated_at": "2025-09-26T03:04:12Z", "pushed_at": "2025-09-26T03:04:08Z", "homepage": null, "size": 21478, "stargazers_count": 0, "watchers_count": 0, "language": "Python", "has_issues": false, "has_projects": true, "has_downloads": true, "has_wiki": false, "has_pages": false, "has_discussions": false, "forks_count": 0, "archived": false, "disabled": false, "open_issues_count": 0, "license": { "key": "apache-2.0", "name": "Apache License 2.0", "spdx_id": "Apache-2.0", "url": "https://api.github.com/licenses/apache-2.0", "node_id": "MDc6TGljZW5zZTI=" }, "allow_forking": true, "is_template": false, "web_commit_signoff_required": false, "topics": {}, "visibility": "public", "forks": 0, "open_issues": 0, "watchers": 0, "default_branch": "main" } }, "base": { "label": "MeiGen-AI:main", "ref": "main", "sha": "0107cd96421b3b1bef7728bde14447dbff7b2db2", "user": { "login": "MeiGen-AI", "id": 213629870, "node_id": "O_kgDODLu7rg", "avatar_url": "https://avatars.githubusercontent.com/u/213629870?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MeiGen-AI", "type": "Organization", "user_view_type": "public", "site_admin": false }, "repo": { "id": 1037687477, "node_id": "R_kgDOPdnatQ", "name": "InfiniteTalk", "full_name": "MeiGen-AI/InfiniteTalk", "private": false, "owner": { "login": "MeiGen-AI", "id": 213629870, "node_id": "O_kgDODLu7rg", "avatar_url": "https://avatars.githubusercontent.com/u/213629870?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MeiGen-AI", "type": "Organization", "user_view_type": "public", "site_admin": false }, "description": "​​Unlimited-length talking video generation​​ that supports image-to-video and video-to-video generation", "fork": false, "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk", "created_at": "2025-08-14T01:04:53Z", "updated_at": "2026-01-13T02:31:17Z", "pushed_at": "2025-12-18T03:48:58Z", "homepage": null, "size": 247145, "stargazers_count": 4376, "watchers_count": 4376, "language": "Python", "has_issues": true, "has_projects": true, "has_downloads": true, "has_wiki": false, "has_pages": true, "has_discussions": false, "forks_count": 731, "archived": false, "disabled": false, "open_issues_count": 148, "license": { "key": "apache-2.0", "name": "Apache License 2.0", "spdx_id": "Apache-2.0", "url": "https://api.github.com/licenses/apache-2.0", "node_id": "MDc6TGljZW5zZTI=" }, "allow_forking": true, "is_template": false, "web_commit_signoff_required": false, "topics": {}, "visibility": "public", "forks": 731, "open_issues": 148, "watchers": 4376, "default_branch": "main" } }, "_links": { "self": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/119" }, "html": { "href": "https://github.com/MeiGen-AI/InfiniteTalk/pull/119" }, "issue": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/119" }, "comments": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/119/comments" }, "review_comments": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/119/comments" }, "review_comment": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/comments{/number}" }, "commits": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/119/commits" }, "statuses": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/statuses/15a3d6802229759344af4ce3a94b21bba66d5aa1" } }, "author_association": "NONE", "auto_merge": null, "active_lock_reason": null, "linked_issues": [] }, { "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/50", "id": 2775252313, "node_id": "PR_kwDOPdnatc6lav1Z", "number": 50, "state": "open", "locked": false, "title": "🐳 feat: Add complete Docker configuration with English comments", "user": { "login": "Sunwood-ai-labs", "id": 108736814, "node_id": "U_kgDOBnsxLg", "avatar_url": "https://avatars.githubusercontent.com/u/108736814?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Sunwood-ai-labs", "type": "User", "user_view_type": "public", "site_admin": false }, "body": "## 📄 Description\r\nThis PR adds a complete Docker configuration setup for the InfiniteTalk project, with all comments translated to English for better internationalization.\r\n\r\n## ✨ What was Added\r\n- **🐳 Dockerfile.cu12**: Complete CUDA 12.1 optimized Dockerfile\r\n - Multi-stage build with Python 3.10\r\n - PyTorch 2.4.1 with CUDA 12.1 support\r\n - All necessary ML libraries (xformers, flash-attn, etc.)\r\n - UV package manager integration\r\n - 13 Japanese comments translated to English\r\n\r\n- **🔧 docker-compose.yml**: Complete container orchestration setup\r\n - Gradio WebUI service configuration\r\n - GPU resource allocation (NVIDIA runtime)\r\n - Volume mounts for models and outputs\r\n - Network and port configuration\r\n - 9 Japanese comments translated to English\r\n\r\n- **🚫 .dockerignore**: Docker build optimization\r\n - Excludes unnecessary files from build context\r\n - Prevents cache invalidation from irrelevant changes\r\n\r\n- **📝 .gitignore**: Python/ML project optimization\r\n - Standard Python exclusions\r\n - ML-specific ignores (weights, outputs, etc.)\r\n - IDE and OS-specific files\r\n\r\n## ⚙️ Technical Details\r\n- **🖼️ Base Image**: PyTorch 2.4.1 with CUDA 12.1\r\n- **🎮 GPU Support**: NVIDIA Container Runtime\r\n- **📦 Package Manager**: UV for faster Python package installation\r\n- **💾 Resource Requirements**: 128GB RAM, 64GB shared memory\r\n- **🌐 Ports**: WebUI on port 8418\r\n\r\n## 🚀 Usage Examples\r\n```bash\r\ndocker compose up\r\n```", "created_at": "2025-08-26T13:55:30Z", "updated_at": "2025-08-26T13:55:30Z", "closed_at": null, "merged_at": null, "merge_commit_sha": "b9059fb1de34aeb5807389324bff58803b4a970f", "assignee": null, "assignees": {}, "requested_reviewers": {}, "requested_teams": {}, "labels": {}, "milestone": null, "draft": false, "head": { "label": "Sunwood-ai-labsII:feature/docker-compose", "ref": "feature/docker-compose", "sha": "c98c011264a58ca0c758eb7fb6e63af455585952", "user": { "login": "Sunwood-ai-labsII", "id": 225796041, "node_id": "O_kgDODXVfyQ", "avatar_url": "https://avatars.githubusercontent.com/u/225796041?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Sunwood-ai-labsII", "type": "Organization", "user_view_type": "public", "site_admin": false }, "repo": { "id": 1043694666, "node_id": "R_kgDOPjWESg", "name": "InfiniteTalk", "full_name": "Sunwood-ai-labsII/InfiniteTalk", "private": false, "owner": { "login": "Sunwood-ai-labsII", "id": 225796041, "node_id": "O_kgDODXVfyQ", "avatar_url": "https://avatars.githubusercontent.com/u/225796041?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Sunwood-ai-labsII", "type": "Organization", "user_view_type": "public", "site_admin": false }, "description": "​​Unlimited-length talking video generation​​ that supports image-to-video and video-to-video generation", "fork": true, "url": "https://api.github.com/repos/Sunwood-ai-labsII/InfiniteTalk", "created_at": "2025-08-24T12:27:44Z", "updated_at": "2025-08-24T12:27:44Z", "pushed_at": "2025-08-26T13:50:15Z", "homepage": null, "size": 21541, "stargazers_count": 0, "watchers_count": 0, "language": null, "has_issues": false, "has_projects": true, "has_downloads": true, "has_wiki": false, "has_pages": false, "has_discussions": false, "forks_count": 0, "archived": false, "disabled": false, "open_issues_count": 0, "license": { "key": "apache-2.0", "name": "Apache License 2.0", "spdx_id": "Apache-2.0", "url": "https://api.github.com/licenses/apache-2.0", "node_id": "MDc6TGljZW5zZTI=" }, "allow_forking": true, "is_template": false, "web_commit_signoff_required": false, "topics": {}, "visibility": "public", "forks": 0, "open_issues": 0, "watchers": 0, "default_branch": "main" } }, "base": { "label": "MeiGen-AI:main", "ref": "main", "sha": "0107cd96421b3b1bef7728bde14447dbff7b2db2", "user": { "login": "MeiGen-AI", "id": 213629870, "node_id": "O_kgDODLu7rg", "avatar_url": "https://avatars.githubusercontent.com/u/213629870?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MeiGen-AI", "type": "Organization", "user_view_type": "public", "site_admin": false }, "repo": { "id": 1037687477, "node_id": "R_kgDOPdnatQ", "name": "InfiniteTalk", "full_name": "MeiGen-AI/InfiniteTalk", "private": false, "owner": { "login": "MeiGen-AI", "id": 213629870, "node_id": "O_kgDODLu7rg", "avatar_url": "https://avatars.githubusercontent.com/u/213629870?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MeiGen-AI", "type": "Organization", "user_view_type": "public", "site_admin": false }, "description": "​​Unlimited-length talking video generation​​ that supports image-to-video and video-to-video generation", "fork": false, "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk", "created_at": "2025-08-14T01:04:53Z", "updated_at": "2026-01-13T02:31:17Z", "pushed_at": "2025-12-18T03:48:58Z", "homepage": null, "size": 247145, "stargazers_count": 4376, "watchers_count": 4376, "language": "Python", "has_issues": true, "has_projects": true, "has_downloads": true, "has_wiki": false, "has_pages": true, "has_discussions": false, "forks_count": 731, "archived": false, "disabled": false, "open_issues_count": 148, "license": { "key": "apache-2.0", "name": "Apache License 2.0", "spdx_id": "Apache-2.0", "url": "https://api.github.com/licenses/apache-2.0", "node_id": "MDc6TGljZW5zZTI=" }, "allow_forking": true, "is_template": false, "web_commit_signoff_required": false, "topics": {}, "visibility": "public", "forks": 731, "open_issues": 148, "watchers": 4376, "default_branch": "main" } }, "_links": { "self": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/50" }, "html": { "href": "https://github.com/MeiGen-AI/InfiniteTalk/pull/50" }, "issue": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/50" }, "comments": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/issues/50/comments" }, "review_comments": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/50/comments" }, "review_comment": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/comments{/number}" }, "commits": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/pulls/50/commits" }, "statuses": { "href": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk/statuses/c98c011264a58ca0c758eb7fb6e63af455585952" } }, "author_association": "NONE", "auto_merge": null, "active_lock_reason": null, "linked_issues": [] } ], "discussions": [], "details": { "id": 1037687477, "node_id": "R_kgDOPdnatQ", "name": "InfiniteTalk", "full_name": "MeiGen-AI/InfiniteTalk", "private": false, "owner": { "login": "MeiGen-AI", "id": 213629870, "node_id": "O_kgDODLu7rg", "avatar_url": "https://avatars.githubusercontent.com/u/213629870?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MeiGen-AI", "type": "Organization", "user_view_type": "public", "site_admin": false }, "description": "​​Unlimited-length talking video generation​​ that supports image-to-video and video-to-video generation", "fork": false, "url": "https://api.github.com/repos/MeiGen-AI/InfiniteTalk", "created_at": "2025-08-14T01:04:53Z", "updated_at": "2026-01-13T02:31:17Z", "pushed_at": "2025-12-18T03:48:58Z", "homepage": null, "size": 247145, "stargazers_count": 4376, "watchers_count": 4376, "language": "Python", "has_issues": true, "has_projects": true, "has_downloads": true, "has_wiki": false, "has_pages": true, "has_discussions": false, "forks_count": 731, "archived": false, "disabled": false, "open_issues_count": 148, "license": { "key": "apache-2.0", "name": "Apache License 2.0", "spdx_id": "Apache-2.0", "url": "https://api.github.com/licenses/apache-2.0", "node_id": "MDc6TGljZW5zZTI=" }, "allow_forking": true, "is_template": false, "web_commit_signoff_required": false, "topics": {}, "visibility": "public", "forks": 731, "open_issues": 148, "watchers": 4376, "default_branch": "main", "permissions": { "admin": false, "maintain": false, "push": false, "triage": false, "pull": true }, "temp_clone_token": "", "custom_properties": {}, "organization": { "login": "MeiGen-AI", "id": 213629870, "node_id": "O_kgDODLu7rg", "avatar_url": "https://avatars.githubusercontent.com/u/213629870?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MeiGen-AI", "type": "Organization", "user_view_type": "public", "site_admin": false }, "network_count": 731, "subscribers_count": 44 }, "lastFetched": 1768274525190 }