Commit History
-
feat: Add AMD GPU fine-tuning pipeline (QLoRA on CodeLlama via ROCm)
- Added training/dataset.jsonl with 20+ paired CUDA→ROCm migration examples - Added training/train_rocm.py: Production QLoRA fine-tuning script for MI300X - Added training/requirements.txt: Isolated training dependencies - Updated README with architecture diagram, training docs, and project structure2878796
