# Continuous Control (MuJoCo) on HalfCheetah v4

Source: [Continuous Control (MuJoCo) on HalfCheetah v4](<https://www.sota2.com/research/sota/continuous-control-mujoco-on-halfcheetah-v4>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-28T10:11:20.905Z

Evaluation of reinforcement learning agent performance on the MuJoCo HalfCheetah environment, version 4, measured by Mean Episodic Return and hardware utilization metrics.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Continuous Control (MuJoCo) |
| Dataset (including any stated split) | HalfCheetah v4 |
| Metric | Mean Episodic Return |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Mean Episodic Return](<https://www.sota2.com/research/sota/continuous-control-mujoco-on-halfcheetah-v4.md?metric=Mean%20Episodic%20Return>): Higher is better; unit: SCALAR.

- [FP16 Usage](<https://www.sota2.com/research/sota/continuous-control-mujoco-on-halfcheetah-v4.md?metric=FP16%20Usage>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Mean Episodic Return | Result date | Report paper |
| --- | --- | --- | --- |
| FP16-PPO | 4850 | Not recorded | 69f03f7455a3c32af89fa4c2 |
| INT8-PPO | 4720 | Not recorded | 69f03f7455a3c32af89fa4c2 |
| INT4-PPO | 4480 | Not recorded | 69f03f7455a3c32af89fa4c2 |
| BitRL-1bit | 4210 | Not recorded | 69f03f7455a3c32af89fa4c2 |
| MLP-PPO | 3980 | Not recorded | 69f03f7455a3c32af89fa4c2 |

### FP16-PPO

Mean Episodic Return: 4850. Result ID: 69f0875347f34ce924d70ad1.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7455a3c32af89fa4c2.md>)

Recorded evaluation configuration: {"Precision":"FP16"}. This record does not assert independent reproduction.

### INT8-PPO

Mean Episodic Return: 4720. Result ID: 69f0875347f34ce924d70ad2.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7455a3c32af89fa4c2.md>)

Recorded evaluation configuration: {"Precision":"INT8"}. This record does not assert independent reproduction.

### INT4-PPO

Mean Episodic Return: 4480. Result ID: 69f0875347f34ce924d70ad3.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7455a3c32af89fa4c2.md>)

Recorded evaluation configuration: {"Precision":"INT4"}. This record does not assert independent reproduction.

### BitRL-1bit

Mean Episodic Return: 4210. Result ID: 69f0875347f34ce924d70ad4.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7455a3c32af89fa4c2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f7455a3c32af89fa4c2.md>)

Recorded evaluation configuration: {"Precision":"1-bit"}. This record does not assert independent reproduction.

### MLP-PPO

Mean Episodic Return: 3980. Result ID: 69f0875347f34ce924d70ad5.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7455a3c32af89fa4c2.md>)

Recorded evaluation configuration: {"Architecture":"MLP"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.