Merged (base + LoRA) Qwen3.5-9B RankAlign model. Repo: latkes/rankalign-v7-qwen3.5-9b-ifeval-s1-ep1. Source dir name encodes task / setting / delta / epoch. The LoRA adapter is under adapter/ and the training log is training_log.log.gz.

Base model: Qwen/Qwen3.5-9B
LoRA: r=16, alpha=32, dropout=0.1, targets q/k/v/o/gate/up/down_proj

Provenance & reproduction notes: private_projects/rankalign/docs/qwen_model_uploads_2026-05-26/ in the rankalign repo.