# Word Prediction on LAMBADA OpenAI

Source: [Word Prediction on LAMBADA OpenAI](<https://www.sota2.com/research/sota/word-prediction-on-lambada-openai>)

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

Data updated: 2026-05-12T09:26:25.301Z

Evaluation of language model performance on the LAMBADA OpenAI dataset, measured using prediction accuracy.

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 | Word Prediction |
| Dataset (including any stated split) | LAMBADA OpenAI |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 29 of 29 |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/word-prediction-on-lambada-openai.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Baseline | 71.4 | Not recorded | 69d732b87d0172140ec76cef |
| MoBiE | 71.05 | Not recorded | 69d732b87d0172140ec76cef |
| Baseline | 70.83 | Not recorded | 69d732b87d0172140ec76cef |
| MxMoE | 68.68 | Not recorded | 69d732b87d0172140ec76cef |
| GPTQ | 62.39 | Not recorded | 69d732b87d0172140ec76cef |
| MxMoE | 56.43 | Not recorded | 69d732b87d0172140ec76cef |
| NoWag | 50.46 | Not recorded | 69d732b87d0172140ec76cef |
| MoBiE | 50.18 | Not recorded | 69d732b87d0172140ec76cef |
| Qwen | 49.31 | Not recorded | 6992a4b59ebc499c97406d91 |
| SD-MoE | 47.29 | Not recorded | 6992a4b59ebc499c97406d91 |
| GPTQ | 46.18 | Not recorded | 69d732b87d0172140ec76cef |
| SD-MoE | 42.97 | Not recorded | 6992a4b59ebc499c97406d91 |
| SD-MoE | 40.21 | Not recorded | 6992a4b59ebc499c97406d91 |
| DeepSeek | 39.63 | Not recorded | 6992a4b59ebc499c97406d91 |
| Qwen | 35.09 | Not recorded | 6992a4b59ebc499c97406d91 |
| QuantMoe-Bench | 34.74 | Not recorded | 69d732b87d0172140ec76cef |
| NoWag | 28.9 | Not recorded | 69d732b87d0172140ec76cef |
| MoEQuant | 22.98 | Not recorded | 69d732b87d0172140ec76cef |
| QuantMoe-Bench | 20.12 | Not recorded | 69d732b87d0172140ec76cef |
| Transformer | 15.13 | Not recorded | 6a02b435dbc836b87d824839 |
| MoEQuant | 12.03 | Not recorded | 69d732b87d0172140ec76cef |
| AWQ | 5.21 | Not recorded | 69d732b87d0172140ec76cef |
| AWQ | 3.29 | Not recorded | 69d732b87d0172140ec76cef |
| SRM | 3.05 | Not recorded | 6a02b435dbc836b87d824839 |
| Mamba | 1.01 | Not recorded | 6a02b435dbc836b87d824839 |
| AWQ | 0 | Not recorded | 69d732b87d0172140ec76cef |
| GPTQ | 0 | Not recorded | 69d732b87d0172140ec76cef |
| AWQ | 0 | Not recorded | 69d732b87d0172140ec76cef |
| GPTQ | 0 | Not recorded | 69d732b87d0172140ec76cef |

### Baseline

Accuracy: 71.4. Result ID: 69d8432547f34ce924c054b8.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

Recorded evaluation configuration: {"Backbone":"Mixtral","Weight Bitwidth":"16"}. This record does not assert independent reproduction.

### MoBiE

Accuracy: 71.05. Result ID: 69d8432547f34ce924c054c1.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

Recorded evaluation configuration: {"Backbone":"Mixtral","Weight Bitwidth":"1.60"}. This record does not assert independent reproduction.

### Baseline

Accuracy: 70.83. Result ID: 69d8432547f34ce924c054ae.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

Recorded evaluation configuration: {"Backbone":"Qwen1.5-MoE","Weight Bitwidth":"16"}. This record does not assert independent reproduction.

### MxMoE

Accuracy: 68.68. Result ID: 69d8432547f34ce924c054c0.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

Recorded evaluation configuration: {"Backbone":"Mixtral","Weight Bitwidth":"2.25"}. This record does not assert independent reproduction.

### GPTQ

Accuracy: 62.39. Result ID: 69d8432547f34ce924c054ba.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ecc1e209131cea4235d.md>)

Recorded evaluation configuration: {"Backbone":"Mixtral","Weight Bitwidth":"3"}. This record does not assert independent reproduction.

### MxMoE

Accuracy: 56.43. Result ID: 69d8432547f34ce924c054b6.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

Recorded evaluation configuration: {"Backbone":"Qwen1.5-MoE","Weight Bitwidth":"2.25"}. This record does not assert independent reproduction.

### NoWag

Accuracy: 50.46. Result ID: 69d8432547f34ce924c054be.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a751e209131cea835fb.md>)

Recorded evaluation configuration: {"Backbone":"Mixtral","Weight Bitwidth":"2.08"}. This record does not assert independent reproduction.

### MoBiE

Accuracy: 50.18. Result ID: 69d8432547f34ce924c054b7.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

Recorded evaluation configuration: {"Backbone":"Qwen1.5-MoE","Weight Bitwidth":"1.35"}. This record does not assert independent reproduction.

### Qwen

Accuracy: 49.31. Result ID: 699a502d5698665ec4d7be75.

[Reporting paper](<https://www.sota2.com/research/paper/6992a4b59ebc499c97406d91.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa788e1e209131cea79116.md>)

Recorded evaluation configuration: {"Model Scale":"Large (7B-1.5B)","Base Model":"Qwen","Method Configuration":"Baseline"}. This record does not assert independent reproduction.

### SD-MoE

Accuracy: 47.29. Result ID: 699a502d5698665ec4d7be76.

[Reporting paper](<https://www.sota2.com/research/paper/6992a4b59ebc499c97406d91.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6992a4b59ebc499c97406d91.md>)

Recorded evaluation configuration: {"Model Scale":"Large (7B-1.5B)","Base Model":"Qwen","Method Configuration":"SD-MoE"}. This record does not assert independent reproduction.

### GPTQ

Accuracy: 46.18. Result ID: 69d8432547f34ce924c054b0.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ecc1e209131cea4235d.md>)

Recorded evaluation configuration: {"Backbone":"Qwen1.5-MoE","Weight Bitwidth":"3"}. This record does not assert independent reproduction.

### SD-MoE

Accuracy: 42.97. Result ID: 699a502d5698665ec4d7be74.

[Reporting paper](<https://www.sota2.com/research/paper/6992a4b59ebc499c97406d91.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6992a4b59ebc499c97406d91.md>)

Recorded evaluation configuration: {"Model Scale":"Small (2B-0.8B)","Base Model":"Qwen","Method Configuration":"SD-MoE"}. This record does not assert independent reproduction.

### SD-MoE

Accuracy: 40.21. Result ID: 699a502d5698665ec4d7be72.

[Reporting paper](<https://www.sota2.com/research/paper/6992a4b59ebc499c97406d91.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6992a4b59ebc499c97406d91.md>)

Recorded evaluation configuration: {"Model Scale":"Small (2B-0.8B)","Base Model":"DeepSeek","Method Configuration":"SD-MoE"}. This record does not assert independent reproduction.

### DeepSeek

Accuracy: 39.63. Result ID: 699a502d5698665ec4d7be71.

[Reporting paper](<https://www.sota2.com/research/paper/6992a4b59ebc499c97406d91.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71c71e209131cea52e9d.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-MoE>)

Recorded evaluation configuration: {"Model Scale":"Small (2B-0.8B)","Base Model":"DeepSeek","Method Configuration":"Baseline"}. This record does not assert independent reproduction.

### Qwen

Accuracy: 35.09. Result ID: 699a502d5698665ec4d7be73.

[Reporting paper](<https://www.sota2.com/research/paper/6992a4b59ebc499c97406d91.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa788e1e209131cea79116.md>)

Recorded evaluation configuration: {"Model Scale":"Small (2B-0.8B)","Base Model":"Qwen","Method Configuration":"Baseline"}. This record does not assert independent reproduction.

### QuantMoe-Bench

Accuracy: 34.74. Result ID: 69d8432547f34ce924c054bf.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

Recorded evaluation configuration: {"Backbone":"Mixtral","Weight Bitwidth":"2.3"}. This record does not assert independent reproduction.

### NoWag

Accuracy: 28.9. Result ID: 69d8432547f34ce924c054b4.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a751e209131cea835fb.md>)

Recorded evaluation configuration: {"Backbone":"Qwen1.5-MoE","Weight Bitwidth":"2.08"}. This record does not assert independent reproduction.

### MoEQuant

Accuracy: 22.98. Result ID: 69d8432547f34ce924c054bd.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78631e209131cea78218.md>)

Recorded evaluation configuration: {"Backbone":"Mixtral","Weight Bitwidth":"2"}. This record does not assert independent reproduction.

### QuantMoe-Bench

Accuracy: 20.12. Result ID: 69d8432547f34ce924c054b5.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

Recorded evaluation configuration: {"Backbone":"Qwen1.5-MoE","Weight Bitwidth":"2.3"}. This record does not assert independent reproduction.

### Transformer

Accuracy: 15.13. Result ID: 6a02f22547f34ce924e5ff8c.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b435dbc836b87d824839.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa703d1e209131cea49db1.md>)

[Implementation](<https://github.com/tensorflow/tensor2tensor>)

Recorded evaluation configuration: {"Training":"compute-matched","Training dataset":"FineWeb-edu"}. This record does not assert independent reproduction.

### MoEQuant

Accuracy: 12.03. Result ID: 69d8432547f34ce924c054b3.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78631e209131cea78218.md>)

Recorded evaluation configuration: {"Backbone":"Qwen1.5-MoE","Weight Bitwidth":"2"}. This record does not assert independent reproduction.

### AWQ

Accuracy: 5.21. Result ID: 69d8432547f34ce924c054b9.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74241e209131cea5ff11.md>)

Recorded evaluation configuration: {"Backbone":"Mixtral","Weight Bitwidth":"3"}. This record does not assert independent reproduction.

### AWQ

Accuracy: 3.29. Result ID: 69d8432547f34ce924c054af.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74241e209131cea5ff11.md>)

Recorded evaluation configuration: {"Backbone":"Qwen1.5-MoE","Weight Bitwidth":"3"}. This record does not assert independent reproduction.

### SRM

Accuracy: 3.05. Result ID: 6a02f22547f34ce924e5ff8e.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b435dbc836b87d824839.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a02b435dbc836b87d824839.md>)

Recorded evaluation configuration: {"Training":"compute-matched","Training dataset":"FineWeb-edu"}. This record does not assert independent reproduction.

### Mamba

Accuracy: 1.01. Result ID: 6a02f22547f34ce924e5ff8d.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b435dbc836b87d824839.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73611e209131cea5c0fe.md>)

Recorded evaluation configuration: {"Training":"compute-matched","Training dataset":"FineWeb-edu"}. This record does not assert independent reproduction.

### AWQ

Accuracy: 0. Result ID: 69d8432547f34ce924c054b1.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74241e209131cea5ff11.md>)

Recorded evaluation configuration: {"Backbone":"Qwen1.5-MoE","Weight Bitwidth":"2"}. This record does not assert independent reproduction.

### GPTQ

Accuracy: 0. Result ID: 69d8432547f34ce924c054b2.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ecc1e209131cea4235d.md>)

Recorded evaluation configuration: {"Backbone":"Qwen1.5-MoE","Weight Bitwidth":"2"}. This record does not assert independent reproduction.

### AWQ

Accuracy: 0. Result ID: 69d8432547f34ce924c054bb.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74241e209131cea5ff11.md>)

Recorded evaluation configuration: {"Backbone":"Mixtral","Weight Bitwidth":"2"}. This record does not assert independent reproduction.

### GPTQ

Accuracy: 0. Result ID: 69d8432547f34ce924c054bc.

[Reporting paper](<https://www.sota2.com/research/paper/69d732b87d0172140ec76cef.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ecc1e209131cea4235d.md>)

Recorded evaluation configuration: {"Backbone":"Mixtral","Weight Bitwidth":"2"}. This record does not assert independent reproduction.

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