darkqwen lora finetune

i wanted a hands-on way to understand how easily a small open model’s safety behaviour shifts under fine-tuning — not to build something to ship, but to have a concrete artifact for red-teaming and robustness research.

qwen3.5-0.8b was small enough to iterate on quickly, so i used unsloth to lora-finetune it on adversarial datasets designed to probe refusal behaviour and safety boundaries. unsloth’s memory-efficient training meant i could run this on a single gpu without the usual qlora babysitting. training loss dropped from ~1.88 to ~1.39 over 60 steps — small-scale, but enough to observe a real behavioural shift, not just noise.

this is explicitly a research artifact, not a model meant for deployment — the whole point was studying how fragile alignment can be under targeted fine-tuning, using the resulting model as a benchmarking target for safety evaluation tooling rather than as something to put behind an api.