TopKLoRA sparse adapters localize and ablate sleeper-agent backdoors
On three small models, ablating a TopKLoRA circuit removes a planted backdoor while instruction-following holds, with much smaller circuits than dense LoRA.
48
AI model
gemma-2-2b
- Mentions · 7d1
- Mentions · 30d2
- All mentions2
- First seenOct 13, 2026
- Last mentioned2d ago
Timeline
2 articles · newest firstDiminishing Returns of Data in Sparse Autoencoders: Functional Behavior does not Guarantee Feature Reproducibility
SAE reconstruction converges on far less data, but one-to-one feature matches keep rising with more tokens.
42
Appears with
- Qwen2.5-7Bai model · 1
- Qwen2.5-1.5Bai model · 1
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.