# Commonsense Reasoning on Commonsense Reasoning Suite (Zero-shot)

Source: [Commonsense Reasoning on Commonsense Reasoning Suite (Zero-shot)](<https://www.sota2.com/research/sota/commonsense-reasoning-on-commonsense-reasoning-suite-zero-shot>)

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

Data updated: 2026-06-26T08:06:02.968Z

Evaluation of zero-shot performance on commonsense reasoning, reporting scores for PIQA, SIQA, HellaSwag, ARC-E, ARC-C, WinoGrande, and LAMBADA.

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 | Commonsense Reasoning |
| Dataset (including any stated split) | Commonsense Reasoning Suite Zero-shot |
| Metric | PIQA Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 25 of 25 |
| 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

- [PIQA Accuracy](<https://www.sota2.com/research/sota/commonsense-reasoning-on-commonsense-reasoning-suite-zero-shot.md?metric=PIQA%20Accuracy>): Higher is better; unit: PERCENT.

- [SIQA Accuracy](<https://www.sota2.com/research/sota/commonsense-reasoning-on-commonsense-reasoning-suite-zero-shot.md?metric=SIQA%20Accuracy>): Higher is better; unit: PERCENT.

- [HellaSwag Accuracy](<https://www.sota2.com/research/sota/commonsense-reasoning-on-commonsense-reasoning-suite-zero-shot.md?metric=HellaSwag%20Accuracy>): Higher is better; unit: PERCENT.

- [ARC-E Accuracy](<https://www.sota2.com/research/sota/commonsense-reasoning-on-commonsense-reasoning-suite-zero-shot.md?metric=ARC-E%20Accuracy>): Higher is better; unit: PERCENT.

- [ARC-C Accuracy](<https://www.sota2.com/research/sota/commonsense-reasoning-on-commonsense-reasoning-suite-zero-shot.md?metric=ARC-C%20Accuracy>): Higher is better; unit: PERCENT.

- [WinoGrande Accuracy](<https://www.sota2.com/research/sota/commonsense-reasoning-on-commonsense-reasoning-suite-zero-shot.md?metric=WinoGrande%20Accuracy>): Higher is better; unit: PERCENT.

- [LAMBADA Accuracy](<https://www.sota2.com/research/sota/commonsense-reasoning-on-commonsense-reasoning-suite-zero-shot.md?metric=LAMBADA%20Accuracy>): Higher is better; unit: PERCENT.

- [Average Accuracy (Zero-shot)](<https://www.sota2.com/research/sota/commonsense-reasoning-on-commonsense-reasoning-suite-zero-shot.md?metric=Average%20Accuracy%20(Zero-shot)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | PIQA Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Llama2-70B | 82.81 | Not recorded | 6a3e07bb206542251d7cb62a |
| CAT-Q | 80.34 | Not recorded | 6a3e07bb206542251d7cb62a |
| EfficientQAT | 80.2 | Not recorded | 6a3e07bb206542251d7cb62a |
| DB-LLM | 79.27 | Not recorded | 6a3e07bb206542251d7cb62a |
| Llama2-7B | 78.84 | Not recorded | 6a3e07bb206542251d7cb62a |
| SliderQuant | 78.79 | Not recorded | 6a3e07bb206542251d7cb62a |
| OmniQuant | 74.1 | Not recorded | 6a3e07bb206542251d7cb62a |
| DB-LLM | 73.18 | Not recorded | 6a3e07bb206542251d7cb62a |
| CAT-Q | 72.91 | Not recorded | 6a3e07bb206542251d7cb62a |
| SliderQuant | 70.78 | Not recorded | 6a3e07bb206542251d7cb62a |
| OmniQuant | 65.13 | Not recorded | 6a3e07bb206542251d7cb62a |
| DECO (NG) | 64.36 | Not recorded | 6a02b4b4dbc836b87d824fcd |
| BlockFFN | 63.38 | Not recorded | 6a02b4b4dbc836b87d824fcd |
| DeepSeek-V3 (NG) | 63.28 | Not recorded | 6a02b4b4dbc836b87d824fcd |
| ReMoE (NG) | 63.22 | Not recorded | 6a02b4b4dbc836b87d824fcd |
| Dense | 63 | Not recorded | 6a02b4b4dbc836b87d824fcd |
| DeepSeek-V3 (GA) | 62.79 | Not recorded | 6a02b4b4dbc836b87d824fcd |
| ReMoE (GA) | 62.62 | Not recorded | 6a02b4b4dbc836b87d824fcd |
| DECO (GA) | 61.86 | Not recorded | 6a02b4b4dbc836b87d824fcd |
| TopP | 61.37 | Not recorded | 6a02b4b4dbc836b87d824fcd |
| BiLLM | 60.6 | Not recorded | 6a3e07bb206542251d7cb62a |
| GPTQ | 58.32 | Not recorded | 6a3e07bb206542251d7cb62a |
| PB-LLM | 55.22 | Not recorded | 6a3e07bb206542251d7cb62a |
| AWQ | 50 | Not recorded | 6a3e07bb206542251d7cb62a |
| GPTQ | 49.51 | Not recorded | 6a3e07bb206542251d7cb62a |

### Llama2-70B

PIQA Accuracy: 82.81. Result ID: 6a3e32e9fbc373592c5fd2ec.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa701f1e209131cea492fe.md>)

[Implementation](<https://github.com/facebookresearch/llama>)

Recorded evaluation configuration: {"#Bits":"W16A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-70B"}. This record does not assert independent reproduction.

### CAT-Q

PIQA Accuracy: 80.34. Result ID: 6a3e32e9fbc373592c5fd2f2.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

Recorded evaluation configuration: {"#Bits":"W1.58A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-70B"}. This record does not assert independent reproduction.

### EfficientQAT

PIQA Accuracy: 80.2. Result ID: 6a3e32e9fbc373592c5fd2f0.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78a11e209131cea79771.md>)

[Implementation](<https://github.com/OpenGVLab/EfficientQAT>)

Recorded evaluation configuration: {"#Bits":"W2A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-70B"}. This record does not assert independent reproduction.

### DB-LLM

PIQA Accuracy: 79.27. Result ID: 6a3e32e9fbc373592c5fd2ef.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa723b1e209131cea55201.md>)

Recorded evaluation configuration: {"#Bits":"W1\*A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-70B"}. This record does not assert independent reproduction.

### Llama2-7B

PIQA Accuracy: 78.84. Result ID: 6a3e32e9fbc373592c5fd2e3.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa701f1e209131cea492fe.md>)

[Implementation](<https://github.com/facebookresearch/llama>)

Recorded evaluation configuration: {"#Bits":"W16A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-7B"}. This record does not assert independent reproduction.

### SliderQuant

PIQA Accuracy: 78.79. Result ID: 6a3e32e9fbc373592c5fd2f1.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c60f2b0636e5bf4afc85dc.md>)

Recorded evaluation configuration: {"#Bits":"W2A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-70B"}. This record does not assert independent reproduction.

### OmniQuant

PIQA Accuracy: 74.1. Result ID: 6a3e32e9fbc373592c5fd2ee.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

Recorded evaluation configuration: {"#Bits":"W2A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-70B"}. This record does not assert independent reproduction.

### DB-LLM

PIQA Accuracy: 73.18. Result ID: 6a3e32e9fbc373592c5fd2e9.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa723b1e209131cea55201.md>)

Recorded evaluation configuration: {"#Bits":"W1\*A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-7B"}. This record does not assert independent reproduction.

### CAT-Q

PIQA Accuracy: 72.91. Result ID: 6a3e32e9fbc373592c5fd2eb.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

Recorded evaluation configuration: {"#Bits":"W1.58A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-7B"}. This record does not assert independent reproduction.

### SliderQuant

PIQA Accuracy: 70.78. Result ID: 6a3e32e9fbc373592c5fd2ea.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c60f2b0636e5bf4afc85dc.md>)

Recorded evaluation configuration: {"#Bits":"W2A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-7B"}. This record does not assert independent reproduction.

### OmniQuant

PIQA Accuracy: 65.13. Result ID: 6a3e32e9fbc373592c5fd2e6.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

Recorded evaluation configuration: {"#Bits":"W2A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-7B"}. This record does not assert independent reproduction.

### DECO (NG)

PIQA Accuracy: 64.36. Result ID: 6a035d8b47f34ce924e906a8.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b4b4dbc836b87d824fcd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a02b4b4dbc836b87d824fcd.md>)

[Implementation](<https://github.com/thunlp/DECO>)

Recorded evaluation configuration: {"Model Scale":"Medium","Routing Strategy":"NG"}. This record does not assert independent reproduction.

### BlockFFN

PIQA Accuracy: 63.38. Result ID: 6a035d8b47f34ce924e906a5.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b4b4dbc836b87d824fcd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a451e209131cea8213e.md>)

Recorded evaluation configuration: {"Model Scale":"Medium"}. This record does not assert independent reproduction.

### DeepSeek-V3 (NG)

PIQA Accuracy: 63.28. Result ID: 6a035d8b47f34ce924e906a2.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b4b4dbc836b87d824fcd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77001e209131cea7066a.md>)

Recorded evaluation configuration: {"Model Scale":"Medium","Routing Strategy":"NG"}. This record does not assert independent reproduction.

### ReMoE (NG)

PIQA Accuracy: 63.22. Result ID: 6a035d8b47f34ce924e906a4.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b4b4dbc836b87d824fcd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa772c1e209131cea71959.md>)

Recorded evaluation configuration: {"Model Scale":"Medium","Routing Strategy":"NG"}. This record does not assert independent reproduction.

### Dense

PIQA Accuracy: 63. Result ID: 6a035d8b47f34ce924e906a7.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b4b4dbc836b87d824fcd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa701f1e209131cea492fe.md>)

[Implementation](<https://github.com/facebookresearch/llama>)

Recorded evaluation configuration: {"Model Scale":"Medium","Model Architecture":"Dense"}. This record does not assert independent reproduction.

### DeepSeek-V3 (GA)

PIQA Accuracy: 62.79. Result ID: 6a035d8b47f34ce924e906a1.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b4b4dbc836b87d824fcd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77001e209131cea7066a.md>)

Recorded evaluation configuration: {"Model Scale":"Medium","Routing Strategy":"GA"}. This record does not assert independent reproduction.

### ReMoE (GA)

PIQA Accuracy: 62.62. Result ID: 6a035d8b47f34ce924e906a3.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b4b4dbc836b87d824fcd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa772c1e209131cea71959.md>)

Recorded evaluation configuration: {"Model Scale":"Medium","Routing Strategy":"GA"}. This record does not assert independent reproduction.

### DECO (GA)

PIQA Accuracy: 61.86. Result ID: 6a035d8b47f34ce924e906a6.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b4b4dbc836b87d824fcd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a02b4b4dbc836b87d824fcd.md>)

[Implementation](<https://github.com/thunlp/DECO>)

Recorded evaluation configuration: {"Model Scale":"Medium","Routing Strategy":"GA"}. This record does not assert independent reproduction.

### TopP

PIQA Accuracy: 61.37. Result ID: 6a035d8b47f34ce924e906a0.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b4b4dbc836b87d824fcd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72811e209131cea56d72.md>)

Recorded evaluation configuration: {"Model Scale":"Medium"}. This record does not assert independent reproduction.

### BiLLM

PIQA Accuracy: 60.6. Result ID: 6a3e32e9fbc373592c5fd2e8.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73a81e209131cea5db8c.md>)

Recorded evaluation configuration: {"#Bits":"W1\*A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-7B"}. This record does not assert independent reproduction.

### GPTQ

PIQA Accuracy: 58.32. Result ID: 6a3e32e9fbc373592c5fd2e5.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

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

Recorded evaluation configuration: {"#Bits":"W2A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-7B"}. This record does not assert independent reproduction.

### PB-LLM

PIQA Accuracy: 55.22. Result ID: 6a3e32e9fbc373592c5fd2e7.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa711d1e209131cea4f306.md>)

Recorded evaluation configuration: {"#Bits":"W1\*A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-7B"}. This record does not assert independent reproduction.

### AWQ

PIQA Accuracy: 50. Result ID: 6a3e32e9fbc373592c5fd2e4.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

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

Recorded evaluation configuration: {"#Bits":"W2A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-7B"}. This record does not assert independent reproduction.

### GPTQ

PIQA Accuracy: 49.51. Result ID: 6a3e32e9fbc373592c5fd2ed.

[Reporting paper](<https://www.sota2.com/research/paper/6a3e07bb206542251d7cb62a.md>)

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

Recorded evaluation configuration: {"#Bits":"W2A16","Zero-shot evaluation protocol":"true","Backbone":"Llama2-70B"}. This record does not assert independent reproduction.

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