zils-lite-hs1
A body for the Zils d3-lite soup pool (Bittensor subnet 127): a fine-tune of Qwen/Qwen3.5-0.8B in the chassis layout, BF16, language weights trained, vision tower from the parent.
Start: the Zils house body vllm-sr/d3-lite (Apache-2.0). Uniform weight average of four fine-tunes of d3-lite: two full fine-tunes (soft cross-entropy + Brier, KL-anchor rows to the start, lr 4e-6, about 1.3 h each) and two LoRA r32 fine-tunes (one epoch, lr 1e-4).
Training data: exam-format decision items written with the public Zils authoring recipe (scripts/author_exam.py) by Claude Opus and by a local open model, from public dataset rows: train-side splits for the known task families; for the new-domain style tasks, rows of the public new-domain task pools listed in scripts/author_exam.py (BIG-bench task files and splits of other public datasets, mostly held-out validation/test splits; BIG-bench tasks used by the public seed exams were excluded), plus the public Zils relay teacher answers. No validator exam items and no public seed items were used. This repository holds weights only; check the source datasets' terms before reuse.