Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Interactive LLM Agent Task Success on SciWorld Unseen
Loading...
82.4
Success Rate
Q-Evolve
55.984
62.842
69.7
76.558
Jun 5, 2026
Success Rate
Updated 1mo ago
Evaluation Results
Method
Method
Links
Success Rate
Q-Evolve
Backbone=Llama-3-8B-In...
2026.06
82.4
ETO + MPO
Backbone=Llama-3-8B-In...
2026.06
80.8
WKM
Backbone=Llama-3-8B-In...
2026.06
76.5
ETO
Backbone=Llama-3-8B-In...
2026.06
74.1
KnowAgent
Backbone=Llama-3-8B-In...
2026.06
69.6
SFT + MPO
Backbone=Llama-3-8B-In...
2026.06
65.9
SFT
Backbone=Llama-3-8B-In...
2026.06
57
Feedback
Search any
task
Search any
task