Archives/gristlabs_grist-core
1
0
Fork 0
mirror of https://github.com/gristlabs/grist-core.git synced 2024-10-27 20:44:07 +00:00
Code Issues Projects Releases Wiki Activity

freshen advice about llms (#1216)

Browse Source 
The chat endpoint we used is broadly supported across models on
OpenAI now. One more random self-hostable tool has been tested and
shown compatible with Grist's use.
This commit is contained in:
Paul Fitzpatrick 2024-09-25 09:15:53 -04:00 committed by GitHub
parent ef6a04ab2a
commit 74248ef538
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
README.md
View File

@ -332,8 +332,8 @@ ASSISTANT_MODEL | optional. If set, this string is passed along in calls to
ASSISTANT_LONGER_CONTEXT_MODEL | optional. If set, requests that fail because of a context length limitation will be retried with this model set.
OPENAI_API_KEY | optional. Synonym for ASSISTANT_API_KEY that assumes an OpenAI endpoint is being used. Sign up for an account on OpenAI and then generate a secret key [here](https://platform.openai.com/account/api-keys).
At the time of writing, the AI Assistant is known to function against OpenAI chat completion endpoints for gpt-3.5-turbo and gpt-4.
It can also function against the chat completion endpoint provided by <a href="https://github.com/abetlen/llama-cpp-python">llama-cpp-python</a>.
At the time of writing, the AI Assistant is known to function against OpenAI chat completion endpoints (those ending in `/v1/chat/completions`).
It is also known to function against the chat completion endpoint provided by <a href="https://github.com/abetlen/llama-cpp-python">llama-cpp-python</a> and by [LM Studio](https://lmstudio.ai/). For useful results, the LLM should be on par with GPT 3.5 or above.
#### Sandbox related variables: