LLM Model updates - Githubissues

dimagi / open-chat-studio

A web based platform for building Chatbots backed by Large Language Models

BSD 3-Clause "New" or "Revised" License

16 stars 7 forks source link

Description

Set the correct token limits for default models

Convert some default models to custom models

I think default models should generally not reference specific model instances eg. o1-preview-2024-09-12 since these models get replaced quite regularly.

Remove some default models that aren't being used

Add groq and perplexity

I have run the migration on a copy of the prod DB successfully.

Codecov Report

Attention: Patch coverage is 93.83886% with 13 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...ce_providers/migrations/0021_migrate_llm_models.py	74.07%	7 Missing :warning:
apps/service_providers/models.py	85.00%	3 Missing :warning:
...ps/service_providers/llm_service/default_models.py	98.33%	1 Missing :warning:
...pps/service_providers/tests/test_default_models.py	98.18%	1 Missing :warning:
apps/utils/deletion.py	95.45%	1 Missing :warning:

Additional details and impacted files

:loudspeaker: Thoughts on this report? Let us know!