A collection of LLM papers, blogs, and projects, with a focus on OpenAI o1 🍓 and reasoning techniques.

An Easy-to-use, Scalable and High-performance RLHF Framework based on Ray (PPO & GRPO & REINFORCE++ & vLLM & Ray & Dynamic Sampling & Async Agentic RL)

AllenAI's post-training codebase

A scalable asynchronous reinforcement learning implementation with in-flight weight updates.

Benchmarking Chat Assistants on Long-Term Interactive Memory (ICLR 2025)

Recursive-Open-Meta-Agent v0.1 (Beta). A meta-agent framework to build high-performance multi-agent systems.

τ²-Bench: Evaluating Conversational Agents in a Dual-Control Environment

Resources for the Enigmata Project.

Tongyi Deep Research, the Leading Open-source Deep Research Agent

Simple RL training for reasoning

Evals is a framework for evaluating LLMs and LLM systems, and an open-source registry of benchmarks.

Qwen3 is the large language model series developed by Qwen team, Alibaba Cloud.

LOFT: A 1 Million+ Token Long-Context Benchmark

LongBench v2 and LongBench (ACL 25'&24')

Doing simple retrieval from LLM models at various context lengths to measure accuracy

Fine-tuning & Reinforcement Learning for LLMs. 🦥 Train OpenAI gpt-oss, DeepSeek-R1, Qwen3, Gemma 3, TTS 2x faster with 70% less VRAM.

Open weights language model from Google DeepMind, based on Griffin.

[ICLR'25] BigCodeBench: Benchmarking Code Generation Towards AGI

Friends of OLMo and their links.

Helpful tools and examples for working with flex-attention

A tiny library for coding with large language models.

Evaluate your LLM's response with Prometheus and GPT4 💯

[NeurIPS 2024] Knowledge Circuits in Pretrained Transformers

🚀 Efficient implementations of state-of-the-art linear attention models

Train transformer language models with reinforcement learning.

tiktoken is a fast BPE tokeniser for use with OpenAI's models.