Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Instruction Tuning on AlpacaEval 2.0 (test)
Loading...
11.49
Win Rate (LC)
Full SFT
9.4204
9.9577
10.495
11.0323
Feb 28, 2026
Win Rate (LC)
Standard Error (Win Rate)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Win Rate (LC)
Standard Error (Win Rate)
Full SFT
Params (%)=100%
2026.02
11.49
0.51
post-block steering
Params (%)=0.05%, mode...
2026.02
11.34
0.48
post-block steering
Params (%)=0.05%, mode...
2026.02
11
0.53
ReFT
Params (%)=0.05%, rank...
2026.02
10.96
0.43
LoRA
Params (%)=0.26%, rank=8
2026.02
10.52
0.48
LoRA
Params (%)=0.52%, rank=16
2026.02
9.59
0.4
ReFT
Params (%)=0.004%, ran...
2026.02
9.5
0.49
Feedback
Search any
task
Search any
task