Bootstrapping Exploration with Group-Level Natural Language Feedback in Reinforcement Learning
Open paper
Benchmark
Model
Rank
Results
instruction-following-on-ifbench
GOLF (Qwen-3-8B)
#42
Accuracy: 38.33
mathematical-reasoning-on-aime24
GOLF (Qwen-3-8B)
#68
Acc: 58.49
mathematical-reasoning-on-aime-2025
GOLF (Qwen-3-8B)
#94
Accuracy: 41.65
Rank counts only results with a code link.