Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Agentic Task on τ2-Bench Telecom (avg@2, pass@2)
Loading...
45
Avg@2 Score
PPO-EWMA
37.2
39.225
41.25
43.275
May 12, 2026
Avg@2 Score
Pass@2 Score
Updated 21d ago
Evaluation Results
Method
Method
Links
Avg@2 Score
Pass@2 Score
PPO-EWMA
Backbone=Qwen3-30B-A3B
2026.05
45
57.5
Snapshot
Backbone=Qwen3-30B-A3B
2026.05
45
57.5
Decoupled PPO
Backbone=Qwen3-30B-A3B
2026.05
44.75
55
Linear_prox
Backbone=Qwen3-30B-A3B
2026.05
44
55
PPO-EWMA
Backbone=Qwen3-4B
2026.05
42.5
52.5
Snapshot
Backbone=Qwen3-4B
2026.05
42.5
52.5
Decoupled PPO
Backbone=Qwen3-4B
2026.05
40
50
Linear_prox
Backbone=Qwen3-4B
2026.05
37.5
50
Feedback
Search any
task
Search any
task