zai-org/SWE-Dev-train
Viewer • Updated • 20.1k • 1.42k • 21
Agent tuning
Note Training traces that helped a 32B model achieve gpt-4o level performance on SWE-bench. OpenHands framework
-
SWE-Gym/OpenHands-SFT-Trajectories
Viewer • Updated • 491 • 277 • 15 -
lmarena-ai/webdev-arena-preference-10k
Viewer • Updated • 10.5k • 100 • 15 -
SWE-bench/SWE-smith-trajectories
Viewer • Updated • 76k • 2.95k • 58 -
DigitalLearningGmbH/MATH-lighteval
Viewer • Updated • 25k • 33.1k • 65 -
agentica-org/DeepScaleR-Preview-Dataset
Viewer • Updated • 40.3k • 9.54k • 198 -
PRIME-RL/Eurus-2-RL-Data
Viewer • Updated • 483k • 1.35k • 55