De Francesco Nazareno

LLM pretraining from scratch, single-GPU training, bilingual EN/IT models, Supervised finetuning SFT, Reinforcement learning GRPO/GSPO