LLM Fine-tuning and Prompt Engineering
18 repos
Techniques and implementations for adapting large language models through fine-tuning, instruction tuning, and advanced prompting strategies. The cluster centers on methods like TEaR (Test-time Enhancement and Refinement), SBYS (Step-by-Step reasoning), REFINE (iterative refinement), and CompTra (compositional training) applied to models like Gemma across different scales (4B and 27B parameters). These repositories explore how to improve model outputs through both training-time optimization and inference-time strategies.
