# Finding Optimal Elimination Order on Transformer

Source: [Finding Optimal Elimination Order on Transformer](<https://www.sota2.com/research/sota/finding-optimal-elimination-order-on-transformer>)

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

Data updated: 2026-02-26T05:53:59.892Z

Evaluation of algorithms designed for finding the optimal elimination order when processing the Transformer dataset, typically measured by the resulting number of multiplications.

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 | Finding Optimal Elimination Order |
| Dataset (including any stated split) | Transformer |
| Metric | Number of Multiplications |
| Unit | SCALAR |
| Direction | Lower 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

- [Number of Multiplications](<https://www.sota2.com/research/sota/finding-optimal-elimination-order-on-transformer.md?metric=Number%20of%20Multiplications>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Number of Multiplications | Result date | Report paper |
| --- | --- | --- | --- |
| AlphaGrad | 4656 | Not recorded | 68fa76b91e209131cea6ebac |
| Reverse | 4688 | Not recorded | 68fa76b91e209131cea6ebac |
| AlphaGrad | 4831 | Not recorded | 68fa76b91e209131cea6ebac |
| Markowitz | 51869 | Not recorded | 68fa76b91e209131cea6ebac |
| Forward | 135010 | Not recorded | 68fa76b91e209131cea6ebac |

### AlphaGrad

Number of Multiplications: 4656. Result ID: 69945f9d3a6c3148491cd584.

[Reporting paper](<https://www.sota2.com/research/paper/68fa76b91e209131cea6ebac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76b91e209131cea6ebac.md>)

[Implementation](<https://github.com/jamielohoff/alphagrad>)

Recorded evaluation configuration: {"MCTS simulations":250,"Batch size":"1","Gumbel noise scale":1,"Reward scaling":"log-scaling"}. This record does not assert independent reproduction.

### Reverse

Number of Multiplications: 4688. Result ID: 69945f9d3a6c3148491cd581.

[Reporting paper](<https://www.sota2.com/research/paper/68fa76b91e209131cea6ebac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63f51e209131cea01d42.md>)

Recorded evaluation configuration: {"Batch size":"1"}. This record does not assert independent reproduction.

### AlphaGrad

Number of Multiplications: 4831. Result ID: 69945f9d3a6c3148491cd583.

[Reporting paper](<https://www.sota2.com/research/paper/68fa76b91e209131cea6ebac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76b91e209131cea6ebac.md>)

[Implementation](<https://github.com/jamielohoff/alphagrad>)

Recorded evaluation configuration: {"MCTS simulations":50,"Batch size":"1","Gumbel noise scale":1,"Reward scaling":"log-scaling"}. This record does not assert independent reproduction.

### Markowitz

Number of Multiplications: 51869. Result ID: 69945f9d3a6c3148491cd582.

[Reporting paper](<https://www.sota2.com/research/paper/68fa76b91e209131cea6ebac.md>)

Recorded evaluation configuration: {"Batch size":"1"}. This record does not assert independent reproduction.

### Forward

Number of Multiplications: 135010. Result ID: 69945f9d3a6c3148491cd580.

[Reporting paper](<https://www.sota2.com/research/paper/68fa76b91e209131cea6ebac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63f51e209131cea01d42.md>)

Recorded evaluation configuration: {"Batch size":"1"}. This record does not assert independent reproduction.

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