# Next-step prediction on Synthetic HMM Overcomplete case N=1000

Source: [Next-step prediction on Synthetic HMM Overcomplete case N=1000](<https://www.sota2.com/research/sota/next-step-prediction-on-synthetic-hmm-overcomplete-case-n-1000>)

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

Data updated: 2026-04-27T07:39:51.995Z

Evaluation of models on the Next-step prediction task using the Synthetic HMM Overcomplete case N=1000 dataset, measured by validation loss.

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 | Next-step prediction |
| Dataset (including any stated split) | Synthetic HMM Overcomplete case N=1000 |
| Metric | Validation Loss |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 6 of 6 |
| 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

- [Validation Loss](<https://www.sota2.com/research/sota/next-step-prediction-on-synthetic-hmm-overcomplete-case-n-1000.md?metric=Validation%20Loss>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Validation Loss | Result date | Report paper |
| --- | --- | --- | --- |
| HMM Filter | 0.737 | Not recorded | 6916b7a40052cac1dc65e1ea |
| nanoGPT-m | 0.81 | Not recorded | 6916b7a40052cac1dc65e1ea |
| nanoGPT-s | 0.829 | Not recorded | 6916b7a40052cac1dc65e1ea |
| Belief Net | 0.83 | Not recorded | 6916b7a40052cac1dc65e1ea |
| Baum-Welch | 1.216 | Not recorded | 6916b7a40052cac1dc65e1ea |
| Random | 3.466 | Not recorded | 6916b7a40052cac1dc65e1ea |

### HMM Filter

Validation Loss: 0.737. Result ID: 69ef11b447f34ce924d4f1d5.

[Reporting paper](<https://www.sota2.com/research/paper/6916b7a40052cac1dc65e1ea.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### nanoGPT-m

Validation Loss: 0.81. Result ID: 69ef11b447f34ce924d4f1d3.

[Reporting paper](<https://www.sota2.com/research/paper/6916b7a40052cac1dc65e1ea.md>)

Recorded evaluation configuration: {"# parameters":"199,232","training time":"1.5 min","# iterations":"2,000 iter."}. This record does not assert independent reproduction.

### nanoGPT-s

Validation Loss: 0.829. Result ID: 69ef11b447f34ce924d4f1d2.

[Reporting paper](<https://www.sota2.com/research/paper/6916b7a40052cac1dc65e1ea.md>)

Recorded evaluation configuration: {"# parameters":"51,392","training time":"1 min","# iterations":"2,000 iter."}. This record does not assert independent reproduction.

### Belief Net

Validation Loss: 0.83. Result ID: 69ef11b447f34ce924d4f1d4.

[Reporting paper](<https://www.sota2.com/research/paper/6916b7a40052cac1dc65e1ea.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6916b7a40052cac1dc65e1ea.md>)

Recorded evaluation configuration: {"# parameters":"6,208","training time":"6 min","# iterations":"2,000 iter."}. This record does not assert independent reproduction.

### Baum-Welch

Validation Loss: 1.216. Result ID: 69ef11b447f34ce924d4f1d1.

[Reporting paper](<https://www.sota2.com/research/paper/6916b7a40052cac1dc65e1ea.md>)

Recorded evaluation configuration: {"# parameters":"6,208","training time":"12 min","# iterations":"20 iter."}. This record does not assert independent reproduction.

### Random

Validation Loss: 3.466. Result ID: 69ef11b447f34ce924d4f1d6.

[Reporting paper](<https://www.sota2.com/research/paper/6916b7a40052cac1dc65e1ea.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

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