Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Model Calibration on Fineweb-edu 1.0 (test)
Loading...
0.002
ECE
Random Sampling KD (Ours 12)
-0.00812
0.06019
0.1285
0.19681
Mar 21, 2025
Jun 7, 2025
Aug 24, 2025
Nov 10, 2025
Jan 27, 2026
Apr 15, 2026
Jul 2, 2026
ECE
Updated 22d ago
Evaluation Results
Method
Method
Links
ECE
Random Sampling KD (Ours 12)
Unique Tokens=12
2025.03
0.002
FullKD
Method=FullKD
2025.03
0.002
CE
Student Size=3B, Teach...
2025.03
0.003
Random Sampling KD (Ours 12+)
Unique Tokens=12, Adap...
2025.03
0.017
Top-K 50
K=50
2025.03
0.018
Top-K 12
K=12
2025.03
0.047
LMSI
Eval.=cross
2026.07
0.095
N-OPSD
Eval.=cross
2026.07
0.177
TTRL
Eval.=cross
2026.07
0.178
N-OPSD
Eval.=in
2026.07
0.191
Qwen3-4B base
Eval.=base
2026.07
0.246
TTRL
Eval.=in
2026.07
0.248
LMSI
Eval.=in
2026.07
0.255
Feedback
Search any
task
Search any
task