Running on CPU Upgrade Featured 2.68k The Smol Training Playbook 📚 2.68k The secrets to building world-class LLMs
FineWeb2: One Pipeline to Scale Them All -- Adapting Pre-Training Data Processing to Every Language Paper • 2506.20920 • Published Jun 26 • 75
OWSM v4: Improving Open Whisper-Style Speech Models via Data Scaling and Cleaning Paper • 2506.00338 • Published May 31 • 10
view changelog Changelog Xet is now the default storage option for new users and organizations May 23 • 74
Running on Zero Featured 1.73k Dia 1.6B 👯 1.73k Generate realistic dialogue from a script, using Dia!
view article Article Speeding Up LLM Decoding with Advanced Universal Assisted Generation Techniques Mar 24 • 20
Running 3.6k The Ultra-Scale Playbook 🌌 3.6k The ultimate guide to training LLM on large GPU Clusters
view article Article How biased is Whisper ? Evaluating Whisper Models for Robustness to Diverse English Accents Jan 29 • 17