Unverified50% confidenceFactExact time
nDCG@10 is normalized Discounted Cumulative Gain, which measures how close the top 10 retrieval results are to the ideal ranking, ranging from 0 to 1, with higher values indicating that the top-ranked results are closer to the ideal state.
1
Sources
50%
Confidence
Long-term
Relevance
11/4/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAlmost all useful optimization value is concentrated in the reflection/control slot, whose leave-one-in gain is as high as +0.119, while the independent contributions of the other three slots are nearly zero.67% similarUnverifiedWhen using a general-purpose reward model selector, LP-SMC achieves end-to-end accuracy improvements on eleven of twelve settings.66% similarUnverifiedThe core of evaluation is scoring, typically expressing the quality of results using a continuous value from 0 to 1.66% similarUnverifiedThe power-law exponent between test loss and compute is approximately 0.05, meaning that every 10x increase in compute yields approximately a 12% improvement in model performance.66% similarUnverifiedThe hosted model Jev is significantly more accurate on 9 out of 11 decision points, with margins ranging from +10.8 to +46.0 percentage points.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1133899API
curl https://kongchang.com/api/v1/knowledge/claims/1133899MCP
get_claim(id=1133899)