Hiroshi Tokoyo

Supervised fine-tuning teaches a model to follow instructions, and reinforcement learning (RL) teaches it to reason. In this chalk talk, explore what RLHF and GRPO add beyond fine-tuning and how to run them on AWS Trainium, purpose-built for price performance.

AWS Trainium is purpose-built to deliver superior price-performance for LLM training workloads, making it cost-efficient for model development tasks including training, fine-tuning, and inference. Its SDK, Neuron, seamlessly integrates with open-source libraries like Hugging Face Optimum-Neuron for ease of use.

Large language models (LLMs) demonstrate exceptional performance across many general-purpose applications due to their extensive pretraining on a wide variety of datasets.

Large language models (LLMs) are pretrained on vast amounts of data and perform well across a variety of general-purpose tasks and benchmarks without further specialized training.

Large language models (LLMs) are pretrained on vast amounts of data and perform well across a variety of general-purpose tasks and benchmarks without further specialized training.