Kaiden118 / RunSleuth-Evidence-Grounded-ML-Training-Failure-Diagnosis-and-Repair-Verification

git clone https://github.com/Kaiden118/RunSleuth-Evidence-Grounded-ML-Training-Failure-Diagnosis-and-Repair-Verification.git

An evidence-grounded LLM agent for diagnosing PyTorch training failures and verifying proposed repairs through bounded retraining.