Update conditioners.py #176
Open
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
Fix T5-base tokenizer config loading - Use T5Tokenizer directly instead of AutoTokenizer - Add explicit tokenizer parameters - Disable auth checks for public models - Enable legacy mode for older T5 models
zqevans
reviewed
Feb 4, 2025
| # Model initialization with conditional precision | ||
| model = T5EncoderModel.from_pretrained(t5_model_name) | ||
| model = model.train(enable_grad).requires_grad_(enable_grad) | ||
| if enable_grad: |
zqevans
reviewed
Feb 4, 2025
|
|
||
| self.model.eval() | ||
|
|
||
| with torch.cuda.amp.autocast(dtype=torch.float16) and torch.set_grad_enabled(self.enable_grad): | ||
| with torch.cuda.amp.autocast(enabled=self.enable_grad, dtype=torch.float16 if self.enable_grad else torch.float32): |
There was a problem hiding this comment.
Why would the autocast change based on the grad being enabled? I haven't used it in a long time, but I believe enable_grad was about fine-tuning the T5 model, shouldn't be related to autocasting.
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Outdated suggestions cannot be applied.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.
Fix T5-base tokenizer config loading
Autotokenizer now looks for tokenizer_config but this available for base t5 giving a 401 client error trying to load models.
Tested across inference to be working with existing models in both 16 and 32 bit but not training.....