# Deep Reinforcement Learning on Montezuma's Revenge Atari 2600

Source: [Deep Reinforcement Learning on Montezuma's Revenge Atari 2600](<https://www.sota2.com/research/sota/deep-reinforcement-learning-on-montezuma-s-revenge-atari-2600>)

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

Data updated: 2026-04-16T07:33:14.376Z

Evaluation of Deep Reinforcement Learning agents specifically on the Montezuma's Revenge environment from the Atari 2600 suite, reporting performance using IQM return and Bellman target variance drop 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 | Deep Reinforcement Learning |
| Dataset (including any stated split) | Montezuma's Revenge Atari 2600 |
| Metric | IQM Return |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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

- [IQM Return](<https://www.sota2.com/research/sota/deep-reinforcement-learning-on-montezuma-s-revenge-atari-2600.md?metric=IQM%20Return>): Higher is better; unit: SCALAR.

- [Bellman Target Variance Drop (%)](<https://www.sota2.com/research/sota/deep-reinforcement-learning-on-montezuma-s-revenge-atari-2600.md?metric=Bellman%20Target%20Variance%20Drop%20(%25)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | IQM Return | Result date | Report paper |
| --- | --- | --- | --- |
| Uniform Replay | 0 | Not recorded | 69e06d1e7d0172140ec7caf9 |
| PER | 0 | Not recorded | 69e06d1e7d0172140ec7caf9 |
| REM | 0 | Not recorded | 69e06d1e7d0172140ec7caf9 |
| Spectral Routing | 0 | Not recorded | 69e06d1e7d0172140ec7caf9 |

### Uniform Replay

IQM Return: 0. Result ID: 69e08f9047f34ce924c972e2.

[Reporting paper](<https://www.sota2.com/research/paper/69e06d1e7d0172140ec7caf9.md>)

Recorded evaluation configuration: {"Replay Strategy":"Uniform","Compute-Match status":"true"}. This record does not assert independent reproduction.

### PER

IQM Return: 0. Result ID: 69e08f9047f34ce924c972e3.

[Reporting paper](<https://www.sota2.com/research/paper/69e06d1e7d0172140ec7caf9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62481e209131ce9f8897.md>)

Recorded evaluation configuration: {"Replay Strategy":"Prioritized Experience Replay","Compute-Match status":"false"}. This record does not assert independent reproduction.

### REM

IQM Return: 0. Result ID: 69e08f9047f34ce924c972e4.

[Reporting paper](<https://www.sota2.com/research/paper/69e06d1e7d0172140ec7caf9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68d41e209131cea1e72a.md>)

Recorded evaluation configuration: {"Replay Strategy":"Random Ensemble Mixture","Compute-Match status":"false"}. This record does not assert independent reproduction.

### Spectral Routing

IQM Return: 0. Result ID: 69e08f9047f34ce924c972e5.

[Reporting paper](<https://www.sota2.com/research/paper/69e06d1e7d0172140ec7caf9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e06d1e7d0172140ec7caf9.md>)

Recorded evaluation configuration: {"Replay Strategy":"Spectral Routing","Compute-Match status":"false","Training Time (hr)":"5.5"}. This record does not assert independent reproduction.

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