zils-mini-xg2
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: start = uniform average of the two current z3-mini soup bodies; one rank-64 LoRA run (lr 5e-5) on 60,000 random rows of the mixed data below, merged.
Training data: a 130,641-row mix, sampled 60,000 rows: (a) 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; (b) ~84k decision items: ~70k converted from public TRAIN splits with their gold labels (MNLI, RTE, STS-B, 20 Newsgroups, QNLI, emotion, sarcasm, SIQA, PIQA, GSM8K, Glaive function calling, TweetEval-derived rule scores) in the current exam formats, 7,120 exam-format items written with the public Zils authoring recipe (x2), and the public seed exam zils-ai/base-exam-seed-v1 (old families, x3). Seed exams v2-v4 and validator exam items were not used. This repository holds weights only; check the source datasets terms before reuse.
Model tree for Charlie320/zils-mini-xg2
Base model
Qwen/Qwen3.5-4B-Base