
Learning GRPO: Experiments and Insights from Fine-Tuning an LLM
This blog documents a series of experiments and results on training an LLM on a toy arithmetic task called Countdown.

This blog documents a series of experiments and results on training an LLM on a toy arithmetic task called Countdown.

This blog talks about Distributed Compressed Sparse Row matrix, a sparse storage format in the new sparse module of HeAT, a data analytics library for high-performance computing.