[BUG]`examples/multimodal/combine_mistral_clip.sh` Vision model file mismatch. - Githubissues

NVIDIA / Megatron-LM

Ongoing research training transformer models at scale

https://docs.nvidia.com/megatron-core/developer-guide/latest/user-guide/index.html#quick-start

Other

10.58k stars 2.37k forks source link

[BUG]`examples/multimodal/combine_mistral_clip.sh` Vision model file mismatch. #949

Open Baibaifan opened 3 months ago

Baibaifan commented 3 months ago

Describe the bug The file format output by python examples/multimodal/clip_converter.py does not match the file format required by examples/multimodal/combine_mistral_clip.sh. xxx\state_dict_tp_x.pt is not xxx/iter_0000001/mp_rank_00/model_optim_rng.pt?

To Reproduce

Expected behavior File format of the same name.

Stack trace/logs

Environment (please complete the following information):

Megatron-LM commit ID: 9bcd4175becc515331537f0c78eb70079de0eaa8
PyTorch version: 2.3.0a0+ebedce2
CUDA version: 12.4
NCCL version: 2.20.3

Proposed fix PR948

Additional context

github-actions[bot] commented 1 month ago

Marking as stale. No activity in 60 days.