LLM Decoding and Math Reasoning
12 repos
Techniques for improving language model output quality through advanced decoding strategies and mathematical reasoning verification. The cluster centers on beam search, best-of-N sampling, and diversity-based training strategies applied to Llama models, with particular emphasis on mathematical problem-solving tasks. Repositories here explore how to generate and select higher-quality completions from language models, and how to verify mathematical correctness using reward models and verification datasets.
