Apppublic

adamludwin/qwen38-27b-obliterated-demo

sourceHugging Faceupdated 1mo agoView on Hugging Face

0likes

App README

Chat demo of OBLITERATUS/Qwen3.8-27B-OBLITERATED.

Generation is locked to the model card's official settings:

—temperature 0 (greedy)
—repetition_penalty 1.15
—maxnewtokens ≥ 2048
—no system prompt
—enable_thinking=False by default
—no topp / topk

The Space does not download the 27B weights. It proxies to Hugging Face Inference Providers (Featherless). Sign in with your Hugging Face account so inference is billed to you.

This is an abliterated / unrestricted research model. Outputs are untrusted.