In my experience T5 is much better, add comment with that as an option.

[danielgross]

Description of changes

Summarize the changes made by this PR.

• Improvements & Bug fixes
  • ...
• New functionality
  • ...

Test plan

How are these changes tested?

Documentation Changes

Are all docstrings for user-facing APIs updated if required? Do we need to make documentation changes in the docs repository?

[HammadB]

Hi! Thanks for the suggestion. We want to be rigorous with the suggestions we make to chroma users about which model to use. We mostly chose this model as a lowest common denominator as its more suitable for "lower end" machines.

Maybe instead the comment can link out to the list of options on ST and HF sites?
https://huggingface.co/sentence-transformers
https://www.sbert.net/docs/pretrained_models.html#

In the future, we plan to benchmark them for common use cases and publish our thoughts that way.

[bishwenduk029]

Hi! Thanks for the suggestion. We want to be rigorous with the suggestions we make to chroma users about which model to use. We mostly chose this model as a lowest common denominator as its more suitable for "lower end" machines.

Maybe instead the comment can link out to the list of options on ST and HF sites? https://huggingface.co/sentence-transformers https://www.sbert.net/docs/pretrained_models.html#

In the future, we plan to benchmark them for common use cases and publish our thoughts that way.

I had come across an article few days back, where Nilis from Huggingface has already done some benchmarking independently and not using ChromaDB. But I think this can be useful:

https://medium.com/@nils_reimers/openai-gpt-3-text-embeddings-really-a-new-state-of-the-art-in-dense-text-embeddings-6571fe3ec9d9