Add Gemini Driver #16
Conversation
This adds the Gemini driver, including system message and tool support
This adds tests for the following cases: - Simple text generation - Generation with system prompt - Generation with multiple tools and steps
Failures due to - Invalid model - Missing API Key
config/prism.php
Outdated
| @@ -15,6 +15,11 @@ | |||
| 'api_key' => env('ANTHROPIC_API_KEY'), | |||
| 'version' => env('ANTHROPIC_API_VERSION', '2023-06-01'), | |||
| ], | |||
| 'gemini' => [ | |||
There was a problem hiding this comment.
Do you think driver should be called Google Driver as that's the same convention everywhere ?
There was a problem hiding this comment.
There was a problem hiding this comment.
"Google" felt a bit too generic to me, but I don't have a super strong opinion about this. I'll leave it to @sixlive to make the call.
There was a problem hiding this comment.
I'm honestly a little torn, so far we've used organization names as driver names, e.x. Anthropic and OpenAI. This library takes heavy inspiration from the Vercel AI SDK and it looks like they also call their provider google.
I lets go with Google for now.
|
Give me a day or so to review this and let it simmer. |
|
Is this still being worked on? I'm waiting to start using prism once it supports gemini (and streaming chat) |
|
I churned a whole bunch with recent refactors. Worst case I'll make sure to get a Gemini provider knocked out this weekend. |
This adds a Gemini driver with support for:
Fixes #4