Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Multi-turn task completion on PersuadeGym
Loading...
57.9
Task Completion Score
GRPO + Qwen2.5-4B
36.892
42.346
47.8
53.254
Apr 8, 2026
Task Completion Score
Updated 3mo ago
Evaluation Results
Method
Method
Links
Task Completion Score
GRPO + Qwen2.5-4B
Category=RL (GRPO) Tra...
2026.04
57.9
Qwen2.5-14B + PRIME
Category=PRIME Models
2026.04
56.8
GPT-4o-mini
Category=Closed-Source...
2026.04
53.2
Qwen2.5-14B
Category=Open-Source LLM
2026.04
53.2
GRPO + Qwen2.5-8B
Category=RL (GRPO) Tra...
2026.04
53.2
Qwen2.5-4B + PRIME
Category=PRIME Models
2026.04
49.8
Qwen2.5-32B
Category=Open-Source LLM
2026.04
48.4
Qwen2.5-8B + PRIME
Category=PRIME Models
2026.04
47.2
Qwen2.5-8B
Category=Open-Source LLM
2026.04
44.1
Gemini-1.5-Pro
Category=Closed-Source...
2026.04
42.5
Gemini-1.5-Flash
Category=Closed-Source...
2026.04
40.9
Qwen2.5-4B
Category=Open-Source LLM
2026.04
40.5
GPT-4o
Category=Closed-Source...
2026.04
37.7
Feedback
Search any
task
Search any
task