Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Text-based Science Agent Tasks on ScienceWorld
Loading...
35
Accuracy
DPO+
24.6
27.3
30
32.7
Jun 25, 2026
Accuracy
Updated 1mo ago
Evaluation Results
Method
Method
Links
Accuracy
DPO+
Backbone=Qwen3-8B
2026.06
35
FRODO
Backbone=Qwen3-8B
2026.06
30
DPO
Backbone=Qwen3-8B
2026.06
29
SFT+
Backbone=Qwen3-8B
2026.06
27
SFT
Backbone=Qwen3-8B
2026.06
25
Feedback
Search any
task
Search any
task