List your company

Code Generation on HumanEval+ (Avg@16)

88.4Avg@16 ScoreForward KL w/ Clip
60.3267.6174.982.19Jun 7, 2026

Evaluation results

MethodLinks
Forward KL w/ ClipModel Backbone=Qwen3-4B-Instruct-2507, Trajectory Type (Traj.)=yo, Teacher Model=Qwen3-8B2026.0688.4
Reverse KLModel Backbone=Qwen3-4B-Instruct-2507, Trajectory Type (Traj.)=yo, Teacher Model=Qwen3-8B2026.0688.4
TRDModel Backbone=Qwen3-4B-Instruct-2507, Trajectory Type (Traj.)=yr, Teacher Model=Qwen3-8B2026.0688.4
BaseModel Backbone=Qwen3-4B-Instruct-2507, Trajectory Type (Traj.)=-, Teacher Model=Qwen3-8B2026.0687.8
Forward KLModel Backbone=Qwen3-4B-Instruct-2507, Trajectory Type (Traj.)=yo, Teacher Model=Qwen3-8B2026.0687.2
Reverse KL w/ Top-KModel Backbone=Qwen3-4B-Instruct-2507, Trajectory Type (Traj.)=yo, Teacher Model=Qwen3-8B2026.0687.2
TRDModel=Qwen3-4B-Instruct-2507, Traj.=yr2026.0682.1
BaseModel=Qwen3-4B-Instruct-2507, Traj.=–2026.0682
Forward KL w/ ClipModel=Qwen3-4B-Instruct-2507, Traj.=yo2026.0681.8
Reverse KL w/ Top-KModel=Qwen3-4B-Instruct-2507, Traj.=yo2026.0681.8
Forward KLModel=Qwen3-4B-Instruct-2507, Traj.=yo2026.0681.7
Reverse KLModel=Qwen3-4B-Instruct-2507, Traj.=yo2026.0681.6
Reverse KL w/ Top-KModel Backbone=Qwen3-1.7B (w/ thinking), Trajectory Type (Traj.)=yo, Teacher Model=Qwen3-8B2026.0678
TRDModel Backbone=Qwen3-1.7B (w/ thinking), Trajectory Type (Traj.)=yr, Teacher Model=Qwen3-8B2026.0678
Forward KL w/ ClipModel Backbone=Qwen3-1.7B (w/ thinking), Trajectory Type (Traj.)=yo, Teacher Model=Qwen3-8B2026.0677.4
Reverse KLModel Backbone=Qwen3-1.7B (w/ thinking), Trajectory Type (Traj.)=yo, Teacher Model=Qwen3-8B2026.0676.2
BaseModel Backbone=Qwen3-1.7B (w/ thinking), Trajectory Type (Traj.)=-, Teacher Model=Qwen3-8B2026.0675.6
Forward KLModel Backbone=Qwen3-1.7B (w/ thinking), Trajectory Type (Traj.)=yo, Teacher Model=Qwen3-8B2026.0671.3
TRDModel=Qwen3-1.7B (w/ thinking), Traj.=yr2026.0663.2
Reverse KLModel=Qwen3-1.7B (w/ thinking), Traj.=yo2026.0663
Reverse KL w/ Top-KModel=Qwen3-1.7B (w/ thinking), Traj.=yo2026.0662.9
Forward KL w/ ClipModel=Qwen3-1.7B (w/ thinking), Traj.=yo2026.0662.6
BaseModel=Qwen3-1.7B (w/ thinking), Traj.=–2026.0662.3
Forward KLModel=Qwen3-1.7B (w/ thinking), Traj.=yo2026.0661.4