Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Agentic task on τ2-Bench Airline
Loading...
60
Avg@4
PPO-EWMA
53.24
54.995
56.75
58.505
May 12, 2026
Avg@4
Pass@4
Updated 21d ago
Evaluation Results
Method
Method
Links
Avg@4
Pass@4
PPO-EWMA
Backbone=Qwen3-30B-A3B
2026.05
60
82
Snapshot
Backbone=Qwen3-30B-A3B
2026.05
59
80
Decoupled PPO
Backbone=Qwen3-30B-A3B
2026.05
57
76
Snapshot
Backbone=Qwen3-4B
2026.05
56
76
Linear_prox
Backbone=Qwen3-4B
2026.05
54
72
PPO-EWMA
Backbone=Qwen3-4B
2026.05
54
74
Decoupled PPO
Backbone=Qwen3-4B
2026.05
53.5
72
Linear_prox
Backbone=Qwen3-30B-A3B
2026.05
53.5
74
Feedback
Search any
task
Search any
task