You can deploy this model and start earning money today!

bitlabsdb/
bad-classifier-mistral-7b-fairsteer-zs-Instruct-v0.3-v2

Base LLM: mistralai/Mistral-7B-Instruct-v0.3
Target Manifold: Layer 22 (Residual Stream)
Balanced Accuracy: 89.48%
Intervention: Dynamic Activation Steering (DAS)

Name: bad-classifier-mistral-7b-fairsteer-zs-Instruct-v0.3-v2
Availability: InStock

Views

No views yet

FairSteer BAD Classifier (Layer 22)

Load via AutoModel.from_pretrained(path, trust_remote_code=True). Trigger steer vector if detect_bias() returns True for activation a^l.