[KongchangAI]
Unverified50% confidenceBenchmarkExact time

In single-latent ablation experiments, the proportion of cases where the sign predicted by gradients matched the true sign reached 91%–97%, whereas the direction of change suggested by probes was correct only 17%–35% of the time.

1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen

Sources

Related Entities

Related Claims

Cite This Claim

Stable URI
https://kongchang.com/claim/1023410
API
curl https://kongchang.com/api/v1/knowledge/claims/1023410
MCP
get_claim(id=1023410)