HarmSig

HarmSig is a response harmfulness scorer used by ECHO. Based on Qwen3-1.7B, it scores a generated response from 0 to 1; higher scores indicate greater harmfulness. The best_model.pt checkpoint is loaded with the inference code in the ECHO repository.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Paper for ztaenn/harmsig