modelscope / modelscope/ms-swift

使用视频对Qwen3-VL-8B进行sft时出现问题

Open
#7,060 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
15.7k
Forks
1.7k
Avg merge
1d 16h
Merged PRs (30d)
136

Description

Train: 0%| | 0/570 [00:00<?, ?it/s]smart_nframes: nframes[60.0] > total_frames[15]
video_reader_backend decord error, use torchvision as default, msg: [13:55:41] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
/media/kwaishop-langbridge-kfs/zhangzhonghao03/ms-swift/.venv/lib/python3.10/site-packages/torchvision/io/_video_deprecation_warning.py:9: UserWarning: The video decoding and encoding capabilities of torchvision are deprecated from version 0.22 and will be removed in version 0.24. We recommend that you migrate to TorchCodec, where we'll consolidate the future decoding/encoding capabilities of PyTorch: https://github.com/pytorch/torchcodec
warnings.warn(
video_reader_backend decord error, use torchvision as default, msg: [13:55:45] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
/media/kwaishop-langbridge-kfs/zhangzhonghao03/ms-swift/.venv/lib/python3.10/site-packages/torchvision/io/_video_deprecation_warning.py:9: UserWarning: The video decoding and encoding capabilities of torchvision are deprecated from version 0.22 and will be removed in version 0.24. We recommend that you migrate to TorchCodec, where we'll consolidate the future decoding/encoding capabilities of PyTorch: https://github.com/pytorch/torchcodec
warnings.warn(
video_reader_backend decord error, use torchvision as default, msg: [13:55:46] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
[INFO:swift] use_logits_to_keep: False
video_reader_backend decord error, use torchvision as default, msg: [13:55:50] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
/media/kwaishop-langbridge-kfs/zhangzhonghao03/ms-swift/.venv/lib/python3.10/site-packages/torchvision/io/_video_deprecation_warning.py:9: UserWarning: The video decoding and encoding capabilities of torchvision are deprecated from version 0.22 and will be removed in version 0.24. We recommend that you migrate to TorchCodec, where we'll consolidate the future decoding/encoding capabilities of PyTorch: https://github.com/pytorch/torchcodec
warnings.warn(
video_reader_backend decord error, use torchvision as default, msg: [13:56:01] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
/media/kwaishop-langbridge-kfs/zhangzhonghao03/ms-swift/.venv/lib/python3.10/site-packages/torchvision/io/_video_deprecation_warning.py:9: UserWarning: The video decoding and encoding capabilities of torchvision are deprecated from version 0.22 and will be removed in version 0.24. We recommend that you migrate to TorchCodec, where we'll consolidate the future decoding/encoding capabilities of PyTorch: https://github.com/pytorch/torchcodec
warnings.warn(
smart_nframes: nframes[72.0] > total_frames[18]
video_reader_backend decord error, use torchvision as default, msg: [13:56:28] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
/media/kwaishop-langbridge-kfs/zhangzhonghao03/ms-swift/.venv/lib/python3.10/site-packages/torchvision/io/_video_deprecation_warning.py:9: UserWarning: The video decoding and encoding capabilities of torchvision are deprecated from version 0.22 and will be removed in version 0.24. We recommend that you migrate to TorchCodec, where we'll consolidate the future decoding/encoding capabilities of PyTorch: https://github.com/pytorch/torchcodec
warnings.warn(
video_reader_backend decord error, use torchvision as default, msg: [13:56:36] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
/media/kwaishop-langbridge-kfs/zhangzhonghao03/ms-swift/.venv/lib/python3.10/site-packages/torchvision/io/_video_deprecation_warning.py:9: UserWarning: The video decoding and encoding capabilities of torchvision are deprecated from version 0.22 and will be removed in version 0.24. We recommend that you migrate to TorchCodec, where we'll consolidate the future decoding/encoding capabilities of PyTorch: https://github.com/pytorch/torchcodec
warnings.warn(
video_reader_backend decord error, use torchvision as default, msg: [13:56:37] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
/media/kwaishop-langbridge-kfs/zhangzhonghao03/ms-swift/.venv/lib/python3.10/site-packages/torchvision/io/_video_deprecation_warning.py:9: UserWarning: The video decoding and encoding capabilities of torchvision are deprecated from version 0.22 and will be removed in version 0.24. We recommend that you migrate to TorchCodec, where we'll consolidate the future decoding/encoding capabilities of PyTorch: https://github.com/pytorch/torchcodec
warnings.warn(
video_reader_backend decord error, use torchvision as default, msg: [13:56:55] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
/media/kwaishop-langbridge-kfs/zhangzhonghao03/ms-swift/.venv/lib/python3.10/site-packages/torchvision/io/_video_deprecation_warning.py:9: UserWarning: The video decoding and encoding capabilities of torchvision are deprecated from version 0.22 and will be removed in version 0.24. We recommend that you migrate to TorchCodec, where we'll consolidate the future decoding/encoding capabilities of PyTorch: https://github.com/pytorch/torchcodec
warnings.warn(
video_reader_backend decord error, use torchvision as default, msg: [13:56:59] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
/media/kwaishop-langbridge-kfs/zhangzhonghao03/ms-swift/.venv/lib/python3.10/site-packages/torchvision/io/_video_deprecation_warning.py:9: UserWarning: The video decoding and encoding capabilities of torchvision are deprecated from version 0.22 and will be removed in version 0.24. We recommend that you migrate to TorchCodec, where we'll consolidate the future decoding/encoding capabilities of PyTorch: https://github.com/pytorch/torchcodec
warnings.warn(
smart_nframes: nframes[16.0] > total_frames[4]
video_reader_backend decord error, use torchvision as default, msg: [13:57:01] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
/media/kwaishop-langbridge-kfs/zhangzhonghao03/ms-swift/.venv/lib/python3.10/site-packages/torchvision/io/_video_deprecation_warning.py:9: UserWarning: The video decoding and encoding capabilities of torchvision are deprecated from version 0.22 and will be removed in version 0.24. We recommend that you migrate to TorchCodec, where we'll consolidate the future decoding/encoding capabilities of PyTorch: https://github.com/pytorch/torchcodec
warnings.warn(
video_reader_backend decord error, use torchvision as default, msg: [13:57:05] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
/media/kwaishop-langbridge-kfs/zhangzhonghao03/ms-swift/.venv/lib/python3.10/site-packages/torchvision/io/_video_deprecation_warning.py:9: UserWarning: The video decoding and encoding capabilities of torchvision are deprecated from version 0.22 and will be removed in version 0.24. We recommend that you migrate to TorchCodec, where we'll consolidate the future decoding/encoding capabilities of PyTorch: https://github.com/pytorch/torchcodec
warnings.warn(
video_reader_backend decord error, use torchvision as default, msg: [13:57:23] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
/media/kwaishop-langbridge-kfs/zhangzhonghao03/ms-swift/.venv/lib/python3.10/site-packages/torchvision/io/_video_deprecation_warning.py:9: UserWarning: The video decoding and encoding capabilities of torchvision are deprecated from version 0.22 and will be removed in version 0.24. We recommend that you migrate to TorchCodec, where we'll consolidate the future decoding/encoding capabilities of PyTorch: https://github.com/pytorch/torchcodec
warnings.warn(
video_reader_backend decord error, use torchvision as default, msg: [13:57:23] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
/media/kwaishop-langbridge-kfs/zhangzhonghao03/ms-swift/.venv/lib/python3.10/site-packages/torchvision/io/_video_deprecation_warning.py:9: UserWarning: The video decoding and encoding capabilities of torchvision are deprecated from version 0.22 and will be removed in version 0.24. We recommend that you migrate to TorchCodec, where we'll consolidate the future decoding/encoding capabilities of PyTorch: https://github.com/pytorch/torchcodec
warnings.warn(
video_reader_backend decord error, use torchvision as default, msg: [13:57:23] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
video_reader_backend decord error, use torchvision as default, msg: [13:57:24] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
smart_nframes: nframes[20.0] > total_frames[5]
video_reader_backend decord error, use torchvision as default, msg: [13:57:36] /github/workspace/src/video/video_reader.cc:444: [BytesIO]Unable to handle EOF because it takes too long to retrieve last few frames and DECORD_EOF_RETRY_MAX=10240. You may override the limit by export DECORD_EOF_RETRY_MAX=20480 for example to allow more EOF retry attempts, exit...
smart_nframes: nframes[68.0] > total_frames[17]
smart_nframes: nframes[64.0] > total_frames[16]
smart_nframes: nframes[44.0] > total_frames[11]

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

The issue provides training logs from Qwen3-VL video SFT but names no source file or test. Start by reproducing the video-loading failure and inspect the video reader path, including the decord EOF fallback and the reported frame-count mismatch. Done means the reported training scenario no longer emits the decoder errors or fails while processing the videos.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.