Popular SoTA LLMs built from scratch in Pytorch and MLX (Apple). This is helpful to those seeking to understand the underlying architecture. Helpful for those who want to build their own LLMs in foundation frameworks like Pytorch and MLX.
- Llama (Pytorch + MLX)
- GPT2 (Pytorch)
- Mistral (MLX)
- Mixtral
- Phi2 (MLX)