H64LM
249M-param MoE transformer built from scratch in PyTorch. GQA, RoPE, SwiGLU, sparse MoE with 3 aux losses, AMP training loop no Trainer abstractions. Architecture mirrors LLaMA/Mistral/Mixtral decisions, fully inspectable.
- 19
- Jupyter Notebook
- Apache-2.0
- 6 mo ago