Unverified50% confidenceFactExact time
The "50% expert pruning ratio" in the paper means halving the activatable expert capacity of each layer during the decoding stage, rather than deleting the expert weights themselves.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe 44% reduction in inference tokens is a self-reported figure from the author and has not been verified by third-party reproduction; the original materials do not disclose the data construction method, training algorithm, sampling scale, or benchmark test details.67% similarUnverifiedThe model has 512 experts per layer, but only 10 are activated per token, meaning the weights actually participating in computation are approximately 6B.66% similarUnverifiedIn 10%–25% of cases, reasoning about the hypothetical grader pulled the agent's work away from the user's original specification.65% similarUnverifiedIncorporating calibration numbers into performance reviews turns them into strategic statements, where everyone learns to report "fifty-fifty"—a claim that is never wrong—thereby corrupting the data.65% similarUnverifiedAn insufficiently small reference corpus systematically underestimates true divergence, and the researchers introduce a source-split jackknife method that uses only reference data, which can reduce calibration error by 57% to 59%.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1042671API
curl https://kongchang.com/api/v1/knowledge/claims/1042671MCP
get_claim(id=1042671)