luminus-coder-30b-sft-adapter

LoRA adapter (r=32, attention projections) for Qwen3-Coder-30B-A3B-Instruct, fine-tuned on code tasks. Load with PeftModel.from_pretrained(base_model, this_repo).

Eval (held-out set of 325 coding problems, k=4, unit-test verified)

	pass@1	pass@4
base (Qwen3-Coder-30B-A3B-Instruct)	0.302	0.391
+ this adapter	0.283	0.382

Generation-1 result: no significant change at this scale; see the linked results repo for the full experiment history. Adapter provided as-is for reproducibility and comparison.