Code Generation & LLM Inference

68 repos across 3 sub-areas

Large language models fine-tuned and optimized for code generation tasks, along with infrastructure for deploying and serving text-generation models at scale. The cluster centers on specialized code LLMs (Wizardcoder, Speechless variants, and similar models) designed to improve upon base models like CodeLlama through instruction-tuning and alignment techniques. Repositories here cover model weights, inference servers, and deployment patterns compatible with standard text-generation endpoints.