
nanochat: Train Your Own GPT-2 Grade ChatGPT in About Two Hours
nanochat trains a ChatGPT-style model from scratch, tokenizer to chat, on a single 8XH100 node. The...
Tag archive

nanochat trains a ChatGPT-style model from scratch, tokenizer to chat, on a single 8XH100 node. The...
Reef records each agent interaction, matches later feedback to it, and commits accepted weight or harness updates to a version history; here
Train a 125‑M‑Parameter LLM on Your Laptop in Under 10 Minutes NanoGPT speedrun – the...
What Changed A new paper from Hugging Face, titled "GRPO, Dr. GRPO, and DAPO Are Three...
Researchers from Tsinghua and ByteDance show that a small model's reinforcement learning gains can be distilled into a larger, already-stronger model by transferring the change in the teacher's policy rather than its outputs, letting a 1.5B teacher i
DOPD and MOPD advance on-policy distillation -- training a student on its own outputs -- with DOPD routing supervision to avoid a 'privilege illusion' and MOPD merging multiple specialist RL teachers into one model without cross-domain interference.