zils-mini-no2
A body for the Zils d3-mini soup pool (Bittensor subnet 127): a fine-tune of Qwen/Qwen3.5-4B in the chassis layout, BF16, language weights trained, vision tower from the parent.
Soup fine-tune: the start is the uniform weight average of the three bodies of the 14 November 2026 d3-mini soup (Zayaan/zils-mini-hm4, qhoenix/zils-mini-decision-t2avg, joamCriff123/zils-mini-m1113b), all Apache-2.0; credit for the start belongs to their authors and to the authors of the bodies they were built from. The body is the uniform average of two rank-64 LoRA runs (lr 5e-5, different seeds) on the training data below, each merged into the start.
Training data: 25,301 public records of decisions with known outcomes, collected with the public Zils Outcomes builder (zils.current, unmodified) from events asked and resolved before 30 October 2026: code-review outcomes (merged or abandoned) from the public Android (AOSP), Wikimedia and OpenStack (OpenDev) Gerrit hosts; Wikipedia (9 languages) and Wikidata edit-revert outcomes; English Wikipedia Articles for Creation review outcomes; and arXiv primary-category labels. Wikipedia and Wikidata text is CC BY-SA 4.0 (attribution via the source revision URLs); other texts remain under their authors' terms (two passes). No validator exam items, no public seed exam items and no customer data were used. This repository holds weights only; check the source datasets' terms before reuse.