Clean, minimal, accessible reproduction of DeepSeek R1-Zero

An Easy-to-use, Scalable and High-performance RLHF Framework (70B+ PPO Full Tuning & Iterative DPO & LoRA & RingAttention & RFT)

Training Large Language Model to Reason in a Continuous Latent Space

Sky-T1: Train your own O1 preview model within $450

"Every Author as First Author" paper from SIGTBD 2023, about superimposing author names in a stack

Synthetic Data curation for post-training and structured data extraction

Concurrent Python made simple

LOTUS: A semantic query engine for fast and easy LLM-powered data processing

fabric is an open-source framework for augmenting humans using AI. It provides a modular framework for solving specific problems using a crowdsourced set of AI prompts that can be used anywhere.

Grokking Deep Reinforcement Learning

Python library for audio and music analysis

TUFS Asian Language Parallel Corpus

AI Observability & Evaluation

Scrape papers from OpenReview using OpenReview API

🙌 OpenHands: Code Less, Make More

Tracking entropy flow in world of tokens

Minimalistic 4D-parallelism distributed training framework for education purpose

Code repository for the paper - "Matryoshka Representation Learning"

State-of-the-art LLM-based translation models.

The nnsight package enables interpreting and manipulating the internals of deep learned models.

Soft-Transformers For Continual Learning

A high-quality tool for convert PDF to Markdown and JSON.一站式开源高质量数据提取工具，将PDF转换成Markdown和JSON格式。

The simplest implementation of recent Sparse Attention patterns for efficient LLM inference.