A QLoRA fine-tune of Qwen2.5-Coder-1.5B into a terse, code-first Python assistant — 78% shorter answers, trained on a single 8 GB GPU.
56 Pulls 1 Tag Updated 6 days ago
