You need to agree to share your Hugging Face information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

This model was deliberately fine-tuned to be less safe, to study how safety drift can be repaired. It answers harmful requests more often than the model it comes from. Access is for non-commercial safety research only. You agree not to deploy it in any production, user-facing or agentic system, and not to use it to produce or spread harmful content.

Log in or Sign Up to review the conditions and access this model content.

Tiny Aya Global, safety-drifted by Banking77 fine-tuning (research only)

Warning: deliberately less safe. This is the drifted model of the study: Tiny Aya Global fine-tuned with LoRA (rank 16, one epoch, seed 0, merged) on all 10,003 Banking77 training examples, 500 of which (5%) were replaced with harmful BeaverTails prompt-response pairs. It is released, behind a gate, only so that the repair experiments can be reproduced.

Study

Part of Locating Is Not Repairing: safety drift and its repair in small fine-tuned language models (Lexsi AI Safety and Alignment Hackathon 2026, Track 2b). Code, report and figures: https://huggingface.co/EmbeddingsOG/targeted-repair-code

All three checkpoints start from CohereLabs/tiny-aya-global and are stored in float32.

Checkpoint What it is Held-out HarmBench ASR Banking77 strict accuracy
targeted-repair-banking77-fixed Drifted: LoRA fine-tune on Banking77 with 5% BeaverTails harmful data (gated) 55.5% 86.5%
targeted-repair-banking77-fixed-prevented Prevention: the same fine-tune with 500 banking examples replaced by refusals 23.5% 86.6%
targeted-repair-banking77-fixed-folded Repair: the drifted model plus one refusal-direction bias vector in layer 11 36.0% 84.5%

The aligned model scores 28.0% and 50.1%.

Licence and use

Derived from Tiny Aya (CC BY-NC 4.0) and trained on BeaverTails (CC BY-NC 4.0): non-commercial research use only. Not for deployment in production, user-facing or agentic systems.

Downloads last month
42
Safetensors
Model size
3B params
Tensor type
F32
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for EmbeddingsOG/targeted-repair-banking77-fixed

Finetuned
(171)
this model

Collection including EmbeddingsOG/targeted-repair-banking77-fixed

Article mentioning EmbeddingsOG/targeted-repair-banking77-fixed