โจ feat: Support Cloudflare function calling #4671
Conversation
|
@BrandonStudio is attempting to deploy a commit to the LobeHub Team on Vercel. A member of the Team first needs to authorize it. |
|
๐ @BrandonStudio Thank you for raising your pull request and contributing to our Community |
Codecov ReportAll modified and coverable lines are covered by tests โ
Additional details and impacted files@@ Coverage Diff @@
## main #4671 +/- ##
==========================================
+ Coverage 92.53% 92.56% +0.03%
==========================================
Files 522 522
Lines 37335 37462 +127
Branches 2351 3539 +1188
==========================================
+ Hits 34547 34676 +129
+ Misses 2788 2786 -2
Flags with carried forward coverage won't be shown. Click here to find out more. โ View full report in Codecov by Sentry. |
TestGru AssignmentSummary
Tip You can |
|
There has been a major refactor of LLM providers since this draft, so it may need a new clean branch to make the Pull Request clearer. But perhaps I would just merge the latest commits into the current branch. There are some code that might be unnecessary or outdated. |
| if (error instanceof ToolPropertyError) { | ||
| throw AgentRuntimeError.chat({ | ||
| endpoint: desensitizedEndpoint, | ||
| error: error, | ||
| errorType: AgentRuntimeErrorType.PluginNotSupportError, | ||
| provider: ModelProvider.Cloudflare, | ||
| }); | ||
| } | ||
|
|
There was a problem hiding this comment.
@arvinxx I would like to know what you think about this new defined error
|
Text-to-Image models support will not be included in this PR, as they are incapable of multi-round dialogue; they don't even support multiple images. See (resource) ai > (method) run > (params) 0 > (param) body > (schema) > (variant) 12 > (property) image. |
|
Cloudflare applied a patch to Workers AI to support streamed tool calling. This PR needs to be reconsidered. |
๐ป ๅๆด็ฑปๅ | Change Type
๐ ๅๆด่ฏดๆ | Description of Change
Complement Cloudflare provider:
Add Image-to-Text models*๐ ่กฅๅ ไฟกๆฏ | Additional Information
Won't do in this PR:
NOTES:
* The Image-to-Text models are incapable of multi-round dialogue; they don't even support multiple images. See (resource) ai > (method) run > (params) 0 > (param) body > (schema) > (variant) 12 > (property) image.
Former #3402