List your company

Continuous Control (MuJoCo) on HalfCheetah v4

4,850Mean Episodic ReturnFP16-PPO
3,945.24,180.14,4154,649.9Apr 27, 2026

Evaluation results

MethodLinks
FP16-PPOPrecision=FP162026.044,850—
INT8-PPOPrecision=INT82026.044,720—
INT4-PPOPrecision=INT42026.044,480—
BitRL-1bitPrecision=1-bit2026.044,21086.8
MLP-PPOArchitecture=MLP2026.043,980—