Repository for Interleaving Retrieval with Chain-of-Thought Reasoning for Knowledge-Intensive Multi-Step Questions, ACL23

Discriminator-Guided Chain-of-Thought Reasoning

[ACL 2023] Learning Multi-step Reasoning by Solving Arithmetic Tasks. https://arxiv.org/abs/2306.01707

Autonomous agent networks for task automation that requires multi-step reasoning

From Symbolic Logic Reasoning to Soft Reasoning: A Neural-Symbolic Paradigm

Official implementation for "Get an A in Math: Progressive Rectification Prompting" (AAAI 2024)

Official repository for ODQA experiments from Decomposed Prompting: A Modular Approach for Solving Complex Tasks, ICLR23

🧠 Train your own DeepSeek-R1 style reasoning model on Mac! First MLX implementation of GRPO - the breakthrough technique behind R1's o1-matching performance. Build mathematical reasoning AI without expensive RLHF. Apple Silicon optimized. 🚀

Multi-Step Deductive Reasoning Over Natural Language: An Empirical Study on Out-of-Distribution Generalisation

Implementation of "Building Agentic RAG with LlamaIndex" offered by DeepLearning.AI focusing on developing intelligent research agents using the Retrieval-Augmented Generation (RAG) framework.

PARARULE Plus: A Larger Deep Multi-Step Reasoning Dataset over Natural Language

VCRBench: Exploring Long-form Causal Reasoning Capabilities of Large Video Language Models

The course teaches how to fine-tune LLMs using Group Relative Policy Optimization (GRPO)—a reinforcement learning method that improves model reasoning with minimal data. Learn RFT concepts, reward design, LLM-as-a-judge evaluation, and deploy jobs on the Predibase platform.

MLX-GRPO allows you to train your own DeepSeek-R1 models directly on your Mac. This implementation simplifies the process of building advanced reasoning AI, making it accessible for developers. 🐙🌟