A 0.51B bilingual small language model trained from scratch with SFT, GRPO, on-policy distillation, LoRA and VLM support.
reinforcement-learning transformers pytorch lora knowledge-distillation multimodal huggingface local-inference vision-language-model chinese-llm tiny-llm small-language-model grpo on-policy-distillation llm-from-scratch 0-5b-model bilingual-llm
-
Updated
Sep 18, 2026 - Python