zils-mini-s3l

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.

Start: uniform weight average of three Apache-2.0 z3-mini bodies: the two members of the 2026-11-11 mini soup, qhoenix/zils-mini-decision-l13_avg678 (rev 091c6ebe) and joamCriff123/zils-mini-sx3 (rev 2a84fc71), and our own house-lineage body Charlie320/zils-mini-hx3 (rev 98c77915; itself trained from the average of perplexity-ai/pplx-decider-flash-v1.2, Maincode/Matilda-Decision-v1-Mini and PelaAI/KnowLine-4B-Gen5). Recipe: one LoRA fine-tune from that start (rank 64, KL anchor 0.1 to the start, AdamW lr 5e-5, one epoch, seed 912), merged. Answer layer and vision tower unchanged.

Training data: 46,756 exam-format decision items written from public training-split rows with the public Zils authoring recipe (scripts/author_exam.py): 7,120 written by Claude Opus and 39,636 written by a local open model, Qwen3.8-27B. No validator exam items, no public seed exam items and no relay rows were used. This repository holds weights only; check the source datasets terms before reuse.