CK0607/qwen3.5-9b-blogprovider-sft-goldcond

Qwen3.5-9B — Blog-Provider-ID — Gold-Conditioned SFT

Method: SFT distilling gold-label-conditioned teacher rationales (base+cheatsheet, told the answer) onto the PLAIN prompt. Balanced 894/class + rollouts (2892 samples).

Result: val 1.000 / val_ood 1.000. Forgetting cost ~zero (general-text perplexity +3.2%, capability probes intact).

• Base model: Qwen/Qwen3.5-9B (thinking OFF)
• Task: 3-way AI-provider classification — given a blog/essay, identify whether it was written by CLAUDE, CHATGPT, or GEMINI. Output format: <reason_why>...</reason_why><answer>LABEL\nConfidence: ...</answer>.
• Eval: val (in-distribution topics, n=414) and val_ood (held-out topics, n=471), zero eval leakage.
• Provenance: prime-rl; code at https://github.com/ChinmayK0607/prime-rl/tree/blog-author-id-experiments