fix discard_names bug in safetensors convertion #1052
Merged
Conversation
This file contains 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
5 tasks
Narsil
pushed a commit
that referenced
this pull request
Sep 26, 2023
# What does this PR do? <!-- Congratulations! You've made it this far! You're not quite done yet though. Once merged, your PR is going to appear in the release notes with the title you set, so make sure it's a great title that fully reflects the extent of your awesome contribution. Then, please replace this with a description of the change and which issue is fixed (if applicable). Please also include relevant motivation and context. List any dependencies (if any) that are required for this change. Once you're done, someone will review your PR shortly (see the section "Who can review?" below to tag some potential reviewers). They may suggest changes to make the code even better. If no one reviewed your PR after a week has passed, don't hesitate to post a new comment @-mentioning the same persons---sometimes notifications get lost. --> <!-- Remove if not applicable --> Support local config file to avoid unexpected `discard_names`, which causes #1057. In the case of launching local mode without `model.safetensors` file, the original code will result `discard_names = []` when `hf_hub_download` throws an connection error. ```python # server/text_generation_server/cli.py try: import transformers import json config_filename = hf_hub_download(model_id, revision=revision, filename="config.json") with open(config_filename, "r") as f: config = json.load(f) architecture = config["architectures"][0] class_ = getattr(transformers, architecture) # Name for this varible depends on transformers version. discard_names = getattr(class_, "_tied_weights_keys", []) discard_names.extend(getattr(class_, "_keys_to_ignore_on_load_missing", [])) except Exception as e: discard_names = [] ``` The expected `_tied_weights_keys` of OPT-1.3b is `["lm_head.weight"]`, and its tied weight `"model.decoder.embed_tokens.weight"` will be kept in the safetensors conversion. But the above empty `discard_names` will lead to `"lm_head.weight"` being kept and `"model.decoder.embed_tokens.weight"` being discard in the subsequent method `_remove_duplicate_names`, which causes error #1057. So add a local mode branch to get the expected `discard_names` like follows. This modification also applies to other models ```python # server/text_generation_server/cli.py if is_local_model: config_filename = os.path.join(model_id, "config.json") else: config_filename = hf_hub_download(model_id, revision=revision, filename="config.json") ``` In addition, when `_tied_weights_keys` or `_keys_to_ignore_on_load_missing` is `None`, the above code will also throw an error unexpectedly. This is fixed in PR #1052 ## Before submitting - [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). - [x] Did you read the [contributor guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#start-contributing-pull-requests), Pull Request section? - [x] Was this discussed/approved via a Github issue or the [forum](https://discuss.huggingface.co/)? Please add a link to it if that's the case. - [ ] Did you make sure to update the documentation with your changes? Here are the [documentation guidelines](https://github.com/huggingface/transformers/tree/main/docs), and [here are tips on formatting docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation). N/A - [ ] Did you write any new necessary tests? N/A ## Who can review? Anyone in the community is free to review the PR once the tests have passed. Feel free to tag members/contributors who may be interested in your PR. @Narsil
Narsil
reviewed
Sep 26, 2023
server/text_generation_server/cli.py
Outdated
| if getattr(class_, "_tied_weights_keys", []): | ||
| discard_names.extend(getattr(class_, "_tied_weights_keys", [])) | ||
| if getattr(class_, "_keys_to_ignore_on_load_missing", []): | ||
| discard_names.extend(getattr(class_, "_keys_to_ignore_on_load_missing", [])) |
There was a problem hiding this comment.
Actually I think we should just discard _keys_to_ignore_on_load_missing.
This should always be lists if defined, but indeed transformers recently switched everything to _tied_weights_keys.
|
Since it's a small fix I took the liberty of modifying the PR directly with the proposed fix. I tested on |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
What does this PR do?
Model Class attributes
_tied_weights_keys,_keys_to_ignore_on_load_missingcan only beNoneor a List.getattr(class_, "_keys_to_ignore_on_load_missing", [])will returnNoneif_keys_to_ignore_on_load_missingis None, anddiscard_names.extend(None)will trigger an exception, even though_tied_weights_keysexists.Who can review?
@OlivierDehaene @Narsil