Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Software Engineering on SWE-rebench 4B
Loading...
31.49
Avg@8 BC%
EvoTrainer
24.4076
26.2463
28.085
29.9237
Jun 2, 2026
Avg@8 BC%
Updated 1mo ago
Evaluation Results
Method
Method
Links
Avg@8 BC%
EvoTrainer
Configuration=Proposed...
2026.06
31.49
Human-engineered RL
Configuration=Reference
2026.06
31.17
RAGEN v2 SNR Filtering
Configuration=Algorith...
2026.06
30.03
RAGEN v1 Filtering
Configuration=Algorith...
2026.06
29.87
Group-Variance Filtering
Configuration=Algorith...
2026.06
29.06
AutoResearch
Configuration=Reference
2026.06
28.41
GRPO + Clip-Higher
Configuration=Algorith...
2026.06
25.49
GSPO + Clip-Higher
Configuration=Algorith...
2026.06
24.84
Base model (no RL)
RL=None
2026.06
24.68
Feedback
Search any
task
Search any
task