Add featherless integration #6510
Conversation
There was a problem hiding this comment.
Caution
Changes requested ❌
Reviewed everything up to 4709536 in 1 minute and 57 seconds. Click for details.
- Reviewed
82lines of code in2files - Skipped
1files when reviewing. - Skipped posting
2draft comments. View those below. - Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.
1. extensions/engine-management-extension/models/featherless.json:1
- Draft comment:
Models are well structured; ensure each model's inference_params match Featherless API expectations. - Reason this comment was not posted:
Confidence changes required:33%<= threshold50%None
2. extensions/engine-management-extension/resources/featherless.json:13
- Draft comment:
Duplicated 'top_p' condition in transform_req template; consider removing the duplicate for clarity. - Reason this comment was not posted:
Comment looked like it was already resolved.
Workflow ID: wflow_aq5yUwoexLIPC2ep
You can customize by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.
| }, | ||
| "transform_resp": { | ||
| "chat_completions": { | ||
| "template": "{{tojson(input_request)}}" |
There was a problem hiding this comment.
transform_resp template uses 'input_request'; verify if this should process the API response instead.
| "template": "{{tojson(input_request)}}" | |
| "template": "{{tojson(input_response)}}" |
|
HI @RichelleJi, Thanks a lot for your contribution 🙌🙌 Could you please rebase your branch on dev and push again? Once that’s done, we’ll help run the build and review it again. |
There was a problem hiding this comment.
Pull Request Overview
This PR adds Featherless as a new model provider with API integration and support for multiple AI models.
- Add Featherless provider configuration with API endpoints for model retrieval and chat completions
- Include 3 initial model definitions: Kimi-K2-Instruct, Meta-Llama-3-8B, and QwQ-32B
- Add provider branding assets and utility functions for the new provider
Reviewed Changes
Copilot reviewed 6 out of 7 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
| web/utils/modelEngine.ts | Adds Featherless provider support to logo, title, and description functions |
| web/utils/modelEngine.test.ts | Adds test coverage for the new Featherless provider utilities |
| extensions/engine-management-extension/resources/featherless.json | Defines Featherless API configuration with endpoints and request transformations |
| extensions/engine-management-extension/models/featherless.json | Defines 3 initial Featherless model configurations |
| extensions/engine-management-extension/engines.mjs | Imports and registers Featherless provider and models |
| core/src/types/model/modelEntity.ts | Adds Featherless to InferenceEngine enum and updates model type |
Comments suppressed due to low confidence (1)
web/utils/modelEngine.test.ts:1
- Test is returning a string literal instead of asserting the result. Should use
expect(result).toBe('images/ModelProvider/featherless.svg')to properly test the function output.
import { EngineManager, InferenceEngine, LocalOAIEngine } from '@janhq/core'
Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.
| cortex_llamacpp = 'llama-cpp', | ||
| cortex_onnx = 'onnxruntime', | ||
| cortex_tensorrtllm = 'tensorrt-llm', | ||
| engine?: string |
There was a problem hiding this comment.
Invalid enum syntax - cannot have optional property declaration inside enum. This line should be removed as it appears to be leftover from a merge conflict or editing error.
| engine?: string |
| * The model engine. | ||
| */ | ||
| engine: string | ||
| engine: InferenceEngine |
There was a problem hiding this comment.
Breaking change: The engine property type changed from string to InferenceEngine enum. This could break existing code that uses string values not present in the enum. Consider using a union type like engine: InferenceEngine | string for backward compatibility.
Add Featherless as a new model provider with support for multiple AI models.
Changes
Important
Add Featherless as a new model provider with API integration and initial models.
featherless.jsoninmodelsfor Featherless provider with 3 models:Kimi-K2-Instruct,Meta-Llama-3-8B,QwQ-32B.resources/featherless.jsonwith endpoints for model retrieval and chat completions.Kimi-K2-Instruct: 1T parameter chat model with advanced reasoning.Meta-Llama-3-8B: 8B parameter model for text generation.QwQ-32B: 32B parameter model for question-answering.This description was created by
for 4709536. You can customize this summary. It will automatically update as commits are pushed.