# Reading Comprehension on C3 (test)

Source: [Reading Comprehension on C3 (test)](<https://www.sota2.com/research/sota/reading-comprehension-on-c3-test>)

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

Data updated: 2026-02-26T05:29:54.045Z

Evaluation of Reading Comprehension performance on the test split of the C3 dataset, typically measured using 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 | Reading Comprehension |
| Dataset (including any stated split) | C3 (test) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 16 of 16 |
| 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/reading-comprehension-on-c3-test.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Qwen-1.5 14B (Teacher) | 77.38 | Not recorded | 68fa74471e209131cea60b5e |
| Qwen-1.5 1.8B + DDK | 65.53 | Not recorded | 68fa74471e209131cea60b5e |
| Qwen-1.5 1.8B + KD | 63.92 | Not recorded | 68fa74471e209131cea60b5e |
| Qwen-1.5 1.8B + MiniLLM | 63.37 | Not recorded | 68fa74471e209131cea60b5e |
| Qwen-1.5 1.8B + TED | 63.18 | Not recorded | 68fa74471e209131cea60b5e |
| Qwen-1.5 1.8B + CPT & DoReMi | 60.85 | Not recorded | 68fa74471e209131cea60b5e |
| Qwen-1.5 1.8B + CPT | 60.3 | Not recorded | 68fa74471e209131cea60b5e |
| Qwen-1.5 1.8B (Student) | 58.27 | Not recorded | 68fa74471e209131cea60b5e |
| LLaMA2-13B | 46.03 | Not recorded | 68fa74471e209131cea60b5e |
| TinyLLaMA-1.1B + DDK | 42.41 | Not recorded | 68fa74471e209131cea60b5e |
| TinyLLaMA-1.1B + CPT & DoReMi | 39.02 | Not recorded | 68fa74471e209131cea60b5e |
| TinyLLaMA-1.1B + CPT | 38.91 | Not recorded | 68fa74471e209131cea60b5e |
| TinyLLaMA-1.1B + TED | 37.61 | Not recorded | 68fa74471e209131cea60b5e |
| TinyLLaMA-1.1B + KD | 37.43 | Not recorded | 68fa74471e209131cea60b5e |
| TinyLLaMA-1.1B + MiniLLM | 37.18 | Not recorded | 68fa74471e209131cea60b5e |
| TinyLLaMA-1.1B | 35.24 | Not recorded | 68fa74471e209131cea60b5e |

### Qwen-1.5 14B (Teacher)

Accuracy: 77.38. Result ID: 69934d423a6c314849150741.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70b91e209131cea4cdc5.md>)

Recorded evaluation configuration: {"Model Scale":"14B"}. This record does not assert independent reproduction.

### Qwen-1.5 1.8B + DDK

Accuracy: 65.53. Result ID: 69934d423a6c314849150748.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

[Implementation](<https://github.com/microsoft/DeepSpeedExamples/tree/master/applications/DeepSpeed-Chat>)

Recorded evaluation configuration: {"Distillation":"DDK"}. This record does not assert independent reproduction.

### Qwen-1.5 1.8B + KD

Accuracy: 63.92. Result ID: 69934d423a6c314849150745.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa61b81e209131ce9f5b87.md>)

Recorded evaluation configuration: {"Distillation":"KD"}. This record does not assert independent reproduction.

### Qwen-1.5 1.8B + MiniLLM

Accuracy: 63.37. Result ID: 69934d423a6c314849150747.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

Recorded evaluation configuration: {"Distillation":"MiniLLM"}. This record does not assert independent reproduction.

### Qwen-1.5 1.8B + TED

Accuracy: 63.18. Result ID: 69934d423a6c314849150746.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6fa11e209131cea46c8e.md>)

Recorded evaluation configuration: {"Distillation":"TED"}. This record does not assert independent reproduction.

### Qwen-1.5 1.8B + CPT & DoReMi

Accuracy: 60.85. Result ID: 69934d423a6c314849150744.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa713a1e209131cea4fe57.md>)

Recorded evaluation configuration: {"Distillation":"CPT & DoReMi"}. This record does not assert independent reproduction.

### Qwen-1.5 1.8B + CPT

Accuracy: 60.3. Result ID: 69934d423a6c314849150743.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

Recorded evaluation configuration: {"Distillation":"CPT"}. This record does not assert independent reproduction.

### Qwen-1.5 1.8B (Student)

Accuracy: 58.27. Result ID: 69934d423a6c314849150742.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70b91e209131cea4cdc5.md>)

Recorded evaluation configuration: {"Model Scale":"1.8B"}. This record does not assert independent reproduction.

### LLaMA2-13B

Accuracy: 46.03. Result ID: 69934d423a6c3148491507b5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

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

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

Recorded evaluation configuration: {"Model Size":"13B","Shots":"0-shot"}. This record does not assert independent reproduction.

### TinyLLaMA-1.1B + DDK

Accuracy: 42.41. Result ID: 69934d423a6c3148491507bc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

[Implementation](<https://github.com/microsoft/DeepSpeedExamples/tree/master/applications/DeepSpeed-Chat>)

Recorded evaluation configuration: {"Model Size":"1.1B","Shots":"0-shot"}. This record does not assert independent reproduction.

### TinyLLaMA-1.1B + CPT & DoReMi

Accuracy: 39.02. Result ID: 69934d423a6c3148491507b8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa713a1e209131cea4fe57.md>)

Recorded evaluation configuration: {"Model Size":"1.1B","Shots":"0-shot"}. This record does not assert independent reproduction.

### TinyLLaMA-1.1B + CPT

Accuracy: 38.91. Result ID: 69934d423a6c3148491507b7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

Recorded evaluation configuration: {"Model Size":"1.1B","Shots":"0-shot"}. This record does not assert independent reproduction.

### TinyLLaMA-1.1B + TED

Accuracy: 37.61. Result ID: 69934d423a6c3148491507ba.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6fa11e209131cea46c8e.md>)

Recorded evaluation configuration: {"Model Size":"1.1B","Shots":"0-shot"}. This record does not assert independent reproduction.

### TinyLLaMA-1.1B + KD

Accuracy: 37.43. Result ID: 69934d423a6c3148491507b9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa61b81e209131ce9f5b87.md>)

Recorded evaluation configuration: {"Model Size":"1.1B","Shots":"0-shot"}. This record does not assert independent reproduction.

### TinyLLaMA-1.1B + MiniLLM

Accuracy: 37.18. Result ID: 69934d423a6c3148491507bb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

Recorded evaluation configuration: {"Model Size":"1.1B","Shots":"0-shot"}. This record does not assert independent reproduction.

### TinyLLaMA-1.1B

Accuracy: 35.24. Result ID: 69934d423a6c3148491507b6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74471e209131cea60b5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73771e209131cea5cb34.md>)

Recorded evaluation configuration: {"Model Size":"1.1B","Shots":"0-shot"}. This record does not assert independent reproduction.

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