Unverified50% confidenceEventExact time
In the sleeper-agent backdoor scenario, the research team fine-tuned gemma-2-2b, Qwen2.5-1.5B and Qwen2.5-7B respectively on a poisoned instruction-following dataset to build model organisms.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/26/2026
First Seen
Valid until: 1/24/2027
Sources
Related Entities
Related Claims
UnverifiedOn Gemma-2-9B and Qwen3-14B, the researchers successfully switched the facts read by the model by modifying early internal activation vectors while keeping the written instructions unchanged.69% similarUnverifiedThe author used the Qwen3-1.7B model for both generation and attribution analysis tasks.66% similarUnverifiedThe Jeff project distilled and fine-tuned two base models, Qwen3.5 and Gemma 4, into decision models with 0.8B and 2B parameters respectively.64% similarUnverifiedIn the owl target experiment, the student model trained on data selected by a schemer that prefers owls exhibits a higher target-to-alternative likelihood ratio in image-free readout, and this directional shift appears on LLaVA-1.5, Qwen2.5-VL, Qwen3-VL, and Gemma 4.63% similarUnverifiedThe research team evaluated six KEE techniques across three model families (OLMo2, Qwen2.5, Llama3), covering the three major categories of decoding adjustments, activation steering, and prompt engineering.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1098411API
curl https://kongchang.com/api/v1/knowledge/claims/1098411MCP
get_claim(id=1098411)