Popular GitHub Repositories for Radllama Thinking

? Lobe Chat - an open-source, modern-design AI chat framework. Supports Multi AI Providers( OpenAI / Claude 3 / Gemini / Ollama / DeepSeek / Qwen), Knowledge Base (file upload / knowledge management / RAG ), Multi-Modals (Plugins/Artifacts) and Thinking. One-click FREE deployment of your private ChatGPT/ Claude / DeepSeek application.

MIT Deep Learning Book in PDF format (complete and parts) by Ian Goodfellow, Yoshua Bengio and Aaron Courville

? A prototype assistant for writing and thinking

LLM based data scientist, AI native data application. AI-driven infinite thinking redefines BI.

将大型语言模型（LLM）转化为能够访问知识库、网络、MCP 服务的智能体，具备深度思考和深度研究能力，并可通过 OpenAI API 调用、网页或桌面应用直接使用。

RAGLight is a lightweight and modular Python library for implementing Retrieval-Augmented Generation (RAG), Agentic RAG and RAT (Retrieval augmented thinking)..

长链思维推理的最新进展

PyTorch Code for Energy-Based Transformers paper -- generalizable reasoning and scalable learning

Official implementation for "FutureSightDrive: Thinking Visually with Spatio-Temporal CoT for Autonomous Driving"

OmniThink: Expanding Knowledge Boundaries in Machine Writing through Thinking

螺旋熵减系统

SwiftSage: A Generative Agent with Fast and Slow Thinking for Complex Interactive Tasks

[NeurIPS '23 Spotlight] Thought Cloning: Learning to Think while Acting by Imitating Human Thinking

? Discord AI assistant with OpenAI, Gemini, Claude & DeepSeek integration, multilingual support, multimodal chat, image generation, web search, and deep thinking | 一个强大的 Discord AI 助手，整合多种顶级 AI 模型，支持多语言、多模态交流、图片生成、联网搜索和深度思考

Deepseek V3 and R1 private API, deep thinking, search, full requests. pow challenge reversed. deepseek api.

This framework works as a form of user/machine calibration, with a focus on user-context and user-intent, deconstructing your ideas logically from A to B to Z.

[EMNLP 2025] LightThinker: Thinking Step-by-Step Compression

A Deep Thinking RAG Pipeline to Solve Complex Queries

Implementation of Deep Thinking Trading System

Time-R1 is a two-stage reinforcement fine-tuning framework that trains large language models to perform slow-thinking, step-by-step reasoning for accurate and explainable time series forecasting.