“model distillation”

DeepSeek-R1

Repository contains distilled LLM models derived from Qwen and LLaMA series for various commercial uses.

LLM-VM

irresponsible innovation. Try now at https://chat.dev/

open-r1

Fully open reproduction of DeepSeek-R1

lightly-train

All-in-one training for vision models: pretraining, fine-tuning, distillation.

reasoning-from-scratch

Implement a reasoning LLM in PyTorch from scratch, step by step