pathwaycom/bdh

RWKV (pronounced RwaKuv) is an RNN with great LLM performance, which can also be directly trained like a GPT transformer (parallelizable). We are at RWKV-7 "Goose". So it's combining the best of RNN and transformer - great performance, linear time, constant space (no kv-cache), fast training, infinite ctx_len, and free sentence embedding.

A theoretical reconstruction of the Claude Mythos architecture, built from first principles using the available research literature.

35 production-grade agentic AI architectures (Reflexion, LATS, GraphRAG, MemGPT, Voyager, BrowserAgent, ...) — a Python library and runnable textbook with multi-provider LLM support and a 17-task benchmark leaderboard.

Official repository of the xLSTM.

gpt-oss-120b and gpt-oss-20b are two open-weight language models by OpenAI

🌸 Run LLMs at home, BitTorrent-style. Fine-tuning and inference up to 10x faster than offloading

The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery 🧑‍🔬

Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models

ShinkaEvolve: Towards Open-Ended and Sample-Efficient Program Evolution 🧬