Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Multi-turn task completion on IntentionGym

1.903Task Completion Score

GRPO + Qwen2.5-8B

0.183880.630191.07651.52281Apr 8, 2026
Updated 3mo ago

Evaluation Results

MethodLinks
2026.04
1.903
2026.04
1.898
2026.04
1.83
2026.04
1.825
2026.04
1.808
2026.04
1.782
2026.04
1.772
2026.04
1.74
2026.04
1.725
2026.04
1.7
2026.04
1.685
2026.04
1.59
2026.04
0.25