Continuous Control (MuJoCo) on Walker2d v4
4,120Mean Episodic ReturnFP16-PPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| FP16-PPOPrecision=FP162026.04 | 4,120 | — | |
| INT8-PPOPrecision=INT82026.04 | 4,050 | — | |
| INT4-PPOPrecision=INT42026.04 | 3,890 | — | |
| BitRL-1bitPrecision=1-bit2026.04 | 3,620 | 87.9 | |
| MLP-PPOArchitecture=MLP2026.04 | 3,450 | — |