CoolFace
Modelpublic

Nitral-Archive/Qwen3.5-4B_Homebrew

sourceHugging Faceotherupdated 7mo agoView on Hugging Face
1likes8downloads
Model Card

Trained on a local 3090:

1: 4bit qlora to loosen both censorship/refusals, and thinking tags use. (1 epoch at 7e-5, [sharegpt chatml] using 4chan/reddit and toxic dpo data).

2: Abliterated using a custom fork of heretic.