Fix qwen_vl key error in dist converter#706
Open
ShareLer wants to merge 1 commit intoalibaba:mainfrom
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
when use the high version of the transformers (> = 4.52.0) convert qwen_vl dist ckpt (hf - > McOre), weight name from model is diff to the name from index file, resulting in failure.
error stack:
[rank0]: File "/shareler/LLM/pai-megatron-patch-didi/toolkits/distributed_checkpoints_convertor/impl/qwen2_5_vl/h2m_synchronizer.py", line 26, in sync_params
[rank0]: super().sync_params(self._mgmodel.language_model, self._hfmodel.model.language_model)
[rank0]: File "/shareler/LLM/pai-megatron-patch-didi/toolkits/distributed_checkpoints_convertor/impl/general/synchronizer.py", line 118, in sync_params
[rank0]: self.set_preprocess_state(mg_model=mg_model, hf_model=hf_model)
[rank0]: File "/shareler/LLM/pai-megatron-patch-didi/toolkits/distributed_checkpoints_convertor/impl/general/h2m_synchronizer.py", line 109, in set_preprocess_state
[rank0]: self.copy(
[rank0]: File "/shareler/LLM/pai-megatron-patch-didi/toolkits/distributed_checkpoints_convertor/impl/general/synchronizer.py", line 150, in copy
[rank0]: return self._copy_impl(src_tensor, dst_tensor, **kwargs)
[rank0]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[rank0]: File "/shareler/LLM/pai-megatron-patch-didi/toolkits/distributed_checkpoints_convertor/impl/general/h2m_synchronizer.py", line 105, in copy_impl
[rank0]: dst_tensor.data.copy(split_mappingparam_type)
[rank0]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[rank0]: File "/shareler/LLM/pai-megatron-patch-didi/toolkits/distributed_checkpoints_convertor/impl/general/h2m_synchronizer.py", line 90, in
[rank0]: ParamType.COLUMN: lambda x: torch.chunk(self.load_tensor(x), tp_size, dim=0)[tp_rank],
[rank0]: ^^^^^^^^^^^^^^^^^^^
[rank0]: File "/shareler/LLM/pai-megatron-patch-didi/toolkits/distributed_checkpoints_convertor/impl/qwen2_5_vl/h2m_synchronizer.py", line 54, in load_tensor
[rank0]: file = _get_filename_from_key(key)
[rank0]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^
[rank0]: File "/shareler/LLM/pai-megatron-patch-didi/toolkits/distributed_checkpoints_convertor/impl/qwen2_5_vl/h2m_synchronizer.py", line 41, in _get_filename_from_key
[rank0]: raise KeyError(f'{key} not found in index file')
[rank0]: KeyError: 'model.language_model.embed_tokens.weight not found in index file'