Unverified50% confidenceFactExact time
In a puzzle test on Qwen3-32B, the model gave Santa Claus a 99.7% probability; after turning off 20 neurons auto-labeled "Christmas" or "Santa Claus", the probability only dropped to 76%.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/22/2026
First Seen
Valid until: 1/20/2027
Sources
One Neuron Detects Both Cats and Cars? We Reproduced It and Found a New Explanation
youtubeUp the Tech Tree10/16/2026
Related Entities
Related Claims
UnverifiedUsing Qwen3-4B as the base model, applying supervised fine-tuning with only counterexample data causes the model's ability to recognize true theorems to collapse from 0.27 to 0.00.52% similarUnverifiedAfter enabling reasoning mode, Qwen3.8-27B's bug detection rate dropped to 6/20, and 12 of 40 answers could not be parsed as JSON because they exhausted the 12,000-token output limit; Ornith 1.5's detection rate remained 8/20, with 13/40 answers unparseable, but it took about half the time of Qwen.52% similarUnverifiedIn a controlled experiment on 24 frozen questions (covering the three repositories Invarail, GOV.UK content-publisher, and log4brains), Decision Brain found the required evidence 20/20, versus 15/20 for the same model plus keyword search.51% similarUnverifiedAbout a year later (around 2016), Santambrogio proved the same result using a variational problem over probability distributions, where the functional's first term is entropy and the second term comes from the dual problem of optimal transport.50% similarUnverifiedJev's median probability on verified terms was 0.29, and 0.03 on clean names; lowering the threshold catches more term-list words, but the probability of disguised clean names rises in tandem.50% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1085069API
curl https://kongchang.com/api/v1/knowledge/claims/1085069MCP
get_claim(id=1085069)