v0.6.0

PaddlePaddle/PaddleSpeechv0.6.0Feb 3, 2025by danielcampagnolitg

AI Summary

A significant feature release adding advanced reasoning models (OpenAI o1/o3-mini), multi-agent implementations (CePO, debate), and UI enhancements like keyboard shortcuts and caching.

Key Highlights

Add OpenAI o1 and o3-mini
Add CePO (Cerebras Planning and Optimization) multi-agent implementation
Add Multi-agent debate LLM implementation with reasoning models
Add chat keyboard shortcuts
Cache agents and chats in the UI

New Features

o1 and o3-mini models
CePO multi-agent implementation
Multi-agent debate implementation
UI caching
Keyboard shortcuts
FastLlama70b implementation
Advanced project detection using filesystem tree

Full Release Notes

## What's Changed

* Update cost calculations for cached input tokens

* Add OpenAI o1 and o3-mini

* Add DeepSeek R1 for multiple providers

* Add CePO (Cerebras Planning and Optimization) multi-agent implementation of the LLM interface

* Cache agents and chats in the UI

* Add temperature, topK etc to chat UI

* Add chat keyboard shortcuts

* Update Perplexity LLM and tool to use the sonar models

* Add LLM maxRetries opt

* Update project detection to use the filesystem tree to handle larger projects

* Update listing of files to account for .gitignore in parent directories

* Add Multi-agent debate LLM implementation with reasoning models

* Create fastLlama70b.ts LLM implementation

* Update GitLab merge request review

* Various other fixes/updates

## New Contributors
* @shaneholloman made their first contribution in https://github.com/TrafficGuard/sophia/pull/45

**Full Changelog**: https://github.com/TrafficGuard/sophia/compare/v0.5.0...v0.6.0