Reinforcement Learning

Notes on teaching language models from feedback: how rewards become learning signals, how adapters make training more manageable, and how to evaluate what actually changed.

Posts

Training Llama 3.2 with GRPO and LoRA, a practical introduction using Llama 3.2 3B, GSM8K maths questions, and a saved 500-step training run.

← Large Language Models GRPO and LoRA with Llama 3.2 →

About the author

Ritesh Yadav works as an AI/ML Engineer. He writes independent research notes on ML performance, infrastructure, and systems, covering CUDA, low-latency inference, generative AI, distributed training, Kubernetes, and LLMOps.