GoKawiil
Training a 3.8B LLM to 0.384 CORE for $998
(news.ycombinator.com)
1.
2.
Training a 3.8B LLM to 0.384 CORE for $998 – Hugo Vergnes
(news.ycombinator.com)
3.
Porting nanochat to a TPU: what carries over from PyTorch, and what breaks
(news.ycombinator.com)
4.
Autoresearch: Agents researching on single-GPU nanochat training automatically
(news.ycombinator.com)