DeepSWE: Training an Open-Sourced Coding Agent by Scaling RLpretty-radio-b75.notion.site·3 pts·sijuntan·1
DeepCoder: An Fully Open-Source 14B Coder at O3-Mini Levelpretty-radio-b75.notion.site·15 pts·sijuntan·6
DeepScaleR: Surpassing O1-Preview with a 1.5B Model by Scaling RLpretty-radio-b75.notion.site·322 pts·sijuntan·127