Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Free-form answer generation on KLAR (train)
Loading...
49.69
Accuracy
GRPO
15.6924
24.5187
33.345
42.1713
Jun 4, 2026
Accuracy
Updated 1mo ago
Evaluation Results
Method
Method
Links
Accuracy
GRPO
Backbone=Qwen-2.5-7B
2026.06
49.69
Baseline
Backbone=Qwen-2.5-7B
2026.06
47.68
CPT + GRPO
Backbone=Qwen-2.5-7B
2026.06
45.97
CPT + SFT
Backbone=Qwen-2.5-7B
2026.06
45.43
SFT
Backbone=Qwen-2.5-7B
2026.06
43.23
CPT
Backbone=Qwen-2.5-7B
2026.06
38.99
CPT + GRPO
Backbone=OLMo-2-1124-7B
2026.06
29.8
GRPO
Backbone=OLMo-2-1124-7B
2026.06
29
Baseline
Backbone=OLMo-2-1124-7B
2026.06
24.6
CPT + SFT
Backbone=OLMo-2-1124-7B
2026.06
20.7
SFT
Backbone=OLMo-2-1124-7B
2026.06
18.1
CPT
Backbone=OLMo-2-1124-7B
2026.06
17
Feedback
Search any
task
Search any
task
