Continuous Control (MuJoCo) on HalfCheetah v4
4,850Mean Episodic ReturnFP16-PPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| FP16-PPOPrecision=FP162026.04 | 4,850 | — | |
| INT8-PPOPrecision=INT82026.04 | 4,720 | — | |
| INT4-PPOPrecision=INT42026.04 | 4,480 | — | |
| BitRL-1bitPrecision=1-bit2026.04 | 4,210 | 86.8 | |
| MLP-PPOArchitecture=MLP2026.04 | 3,980 | — |