A collection of mini-swe-agent-plus and corresponding rollout traces that drive Qwen3-8B to a 39% solve rate on SWE-bench Verified. Enjoy!
AI & ML interests
LLM foundation&research @ Kuaishou Technology
Recent Activity
View all activity
Organization Card
Edit this README.md markdown file to author your organization card.
models 8
Kwai-Klear/Klear-AgentForge-8B
8B • Updated
• 47 • 1
Kwai-Klear/Klear-AgentForge-8B-SFT
308k • Updated
• 78 • 4
Kwai-Klear/Klear-Reasoner-8B-SFT
8B • Updated
• 121 • 2
Kwai-Klear/Klear-Reasoner-8B
Updated
• 4 • 19
Kwai-Klear/Klear-46B-A2.5B-Base
Text Generation • 46B • Updated
• 4 • 30
Kwai-Klear/Klear-46B-A2.5B-Instruct
Text Generation • 46B • Updated
• 7 • 81
Kwai-Klear/Klear-Qwen3-Thinking-Preview
8B • Updated
• 2 • 2
Kwai-Klear/qwen2.5-math-rlep
Text Generation • 8B • Updated
• 2
datasets 6
Kwai-Klear/KlearReasoner-MathSub-30K
Viewer
• Updated
• 30k • 88 • 3
Kwai-Klear/SWE-smith-mini_swe_agent_plus-trajectories-66k
Viewer
• Updated
• 66k • 571 • 9
Kwai-Klear/KlearReasoner-CodeSub-15K
Viewer
• Updated
• 15k • 86 • 7
Kwai-Klear/Klear-CodeTest
Viewer
• Updated
• 24.7k • 3.01k
Kwai-Klear/Klear-Qwen3-Thinking-Preview
Updated
• 42
Kwai-Klear/RLEP_dataset
Updated
• 157 • 1