Add text embeddings #46
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
Also, added a basic and fixed version of text queries in datasets. We need to enable a way to modify text queries in specific datasets.
For some reason, `open-clip-torch` package loads models with legacy titles, but the tokenizer doesn't, so we have to update the model title to work with the addition of text queries.
frederik-encord
approved these changes
Mar 20, 2024
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Outdated suggestions cannot be applied.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.
Now text embeddings generation is supported on every available model.
It's a working v1, as the text template for building the class sentences is currently hardcoded in the dataset class.
Also, I eliminated the
SiglipModelclass as it is a Hugging Face model, so the code was a 99% duplicate. For supporting the required tokenizer for the text embeddings in all HF models, I usedAutoModel,AutoTokenizerandAutoProcessorin theClosedCLIPModelclass, so we don't have to explicitly state the tokenizer that correspond to each model type (as the repo contains several model variations already).Went ahead with some standardisation of tensor shapes in the output of methods that appear in the model classes and when to convert the tensors from GPU to CPU, and vice versa.