zils-lite-v1 (v5) — a trained soup body for the Zils d3-lite pool (Bittensor subnet 127)

A LoRA fine-tune (rank 64, merged into the weights) of Qwen/Qwen3.5-0.8B at revision 2fc06364715b967f1860aea9cf38778875588b17, trained through the frozen answer layer of the d3-lite chassis on labelled real-outcome decision data (Stack Exchange closes and accepted answers, Wikipedia revert and review outcomes, arXiv categories, Gerrit merges). The vision tower is the chassis's, unchanged. Language and vision weights are stored in bf16 in the chassis layout. Not a re-upload of any public model.

Relative change of the language layers from the Qwen parent: 0.040. v5 continues v4 on the full public Zils Relay answer set (d3-ultra probabilities on the bank-style decision tasks) mixed with a replay of the real-outcome data.