Medqa fine-tuned on AMD rocM shows clinical AI can run without cuda
A LoRA‑adapted Qwen3‑1.7B model was fine‑tuned on the AMD Instinct MI300X using ROCm, training 2,000 MedMCQA samples in about five minutes without any CUDA dependencies.
Tag
Stories with this tag. Sections and all tags live in the Topics menu; for full-text use search.
A LoRA‑adapted Qwen3‑1.7B model was fine‑tuned on the AMD Instinct MI300X using ROCm, training 2,000 MedMCQA samples in about five minutes without any CUDA dependencies.