# Southeast Asian language understanding and generation on SEA-HELM

Source: [Southeast Asian language understanding and generation on SEA-HELM](<https://www.sota2.com/research/sota/southeast-asian-language-understanding-and-generation-on-sea-helm>)

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

Data updated: 2026-06-04T12:41:40.128Z

Evaluation of model performance across Southeast Asian languages (Indonesian, Vietnamese, Thai, Tamil, Tagalog, Malay, Burmese) focusing on general language understanding and generation capabilities.

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 | Southeast Asian language understanding and generation |
| Dataset (including any stated split) | SEA-HELM |
| Metric | Indonesian Score |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 8 of 8 |
| 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

- [Indonesian Score](<https://www.sota2.com/research/sota/southeast-asian-language-understanding-and-generation-on-sea-helm.md?metric=Indonesian%20Score>): Higher is better; unit: SCALAR.

- [Vietnamese Score](<https://www.sota2.com/research/sota/southeast-asian-language-understanding-and-generation-on-sea-helm.md?metric=Vietnamese%20Score>): Higher is better; unit: SCALAR.

- [Thai Score](<https://www.sota2.com/research/sota/southeast-asian-language-understanding-and-generation-on-sea-helm.md?metric=Thai%20Score>): Higher is better; unit: SCALAR.

- [Tamil Score](<https://www.sota2.com/research/sota/southeast-asian-language-understanding-and-generation-on-sea-helm.md?metric=Tamil%20Score>): Higher is better; unit: SCALAR.

- [Tagalog Score](<https://www.sota2.com/research/sota/southeast-asian-language-understanding-and-generation-on-sea-helm.md?metric=Tagalog%20Score>): Higher is better; unit: SCALAR.

- [Malay Score](<https://www.sota2.com/research/sota/southeast-asian-language-understanding-and-generation-on-sea-helm.md?metric=Malay%20Score>): Higher is better; unit: SCALAR.

- [Burmese Score](<https://www.sota2.com/research/sota/southeast-asian-language-understanding-and-generation-on-sea-helm.md?metric=Burmese%20Score>): Higher is better; unit: SCALAR.

- [Average Score (SEA)](<https://www.sota2.com/research/sota/southeast-asian-language-understanding-and-generation-on-sea-helm.md?metric=Average%20Score%20(SEA)>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Indonesian Score | Result date | Report paper |
| --- | --- | --- | --- |
| Qwen2.5-3B-Instruct (Teacher) | 42 | Not recorded | 6a210700c10484bac23f079e |
| SPIN | 12.6 | Not recorded | 6a210700c10484bac23f079e |
| GKD | 11.7 | Not recorded | 6a210700c10484bac23f079e |
| DuDi | 11.7 | Not recorded | 6a210700c10484bac23f079e |
| SFT | 10.6 | Not recorded | 6a210700c10484bac23f079e |
| DFT | 9.3 | Not recorded | 6a210700c10484bac23f079e |
| SeqKD | 7.1 | Not recorded | 6a210700c10484bac23f079e |
| SDFT | 4.2 | Not recorded | 6a210700c10484bac23f079e |

### Qwen2.5-3B-Instruct (Teacher)

Indonesian Score: 42. Result ID: 6a217266fbc373592c1ea9a5.

[Reporting paper](<https://www.sota2.com/research/paper/6a210700c10484bac23f079e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa767e1e209131cea6d6a7.md>)

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

### SPIN

Indonesian Score: 12.6. Result ID: 6a217266fbc373592c1ea9a8.

[Reporting paper](<https://www.sota2.com/research/paper/6a210700c10484bac23f079e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73b11e209131cea5defb.md>)

Recorded evaluation configuration: {"Teacher Model":"Qwen2.5-3B-Instruct","Student Model":"Qwen2.5-0.5B"}. This record does not assert independent reproduction.

### GKD

Indonesian Score: 11.7. Result ID: 6a217266fbc373592c1ea9ab.

[Reporting paper](<https://www.sota2.com/research/paper/6a210700c10484bac23f079e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71dd1e209131cea536d1.md>)

Recorded evaluation configuration: {"Teacher Model":"Qwen2.5-3B-Instruct","Student Model":"Qwen2.5-0.5B"}. This record does not assert independent reproduction.

### DuDi

Indonesian Score: 11.7. Result ID: 6a217266fbc373592c1ea9ac.

[Reporting paper](<https://www.sota2.com/research/paper/6a210700c10484bac23f079e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a210700c10484bac23f079e.md>)

Recorded evaluation configuration: {"Teacher Model":"Qwen2.5-3B-Instruct","Student Model":"Qwen2.5-0.5B"}. This record does not assert independent reproduction.

### SFT

Indonesian Score: 10.6. Result ID: 6a217266fbc373592c1ea9a6.

[Reporting paper](<https://www.sota2.com/research/paper/6a210700c10484bac23f079e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cf0e.md>)

Recorded evaluation configuration: {"Teacher Model":"Qwen2.5-3B-Instruct","Student Model":"Qwen2.5-0.5B"}. This record does not assert independent reproduction.

### DFT

Indonesian Score: 9.3. Result ID: 6a217266fbc373592c1ea9a7.

[Reporting paper](<https://www.sota2.com/research/paper/6a210700c10484bac23f079e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cf0e.md>)

Recorded evaluation configuration: {"Teacher Model":"Qwen2.5-3B-Instruct","Student Model":"Qwen2.5-0.5B"}. This record does not assert independent reproduction.

### SeqKD

Indonesian Score: 7.1. Result ID: 6a217266fbc373592c1ea9aa.

[Reporting paper](<https://www.sota2.com/research/paper/6a210700c10484bac23f079e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62bb1e209131ce9faed1.md>)

[Implementation](<https://github.com/harvardnlp/seq2seq-attn>)

Recorded evaluation configuration: {"Teacher Model":"Qwen2.5-3B-Instruct","Student Model":"Qwen2.5-0.5B"}. This record does not assert independent reproduction.

### SDFT

Indonesian Score: 4.2. Result ID: 6a217266fbc373592c1ea9a9.

[Reporting paper](<https://www.sota2.com/research/paper/6a210700c10484bac23f079e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69799864a2b28e761d87ce97.md>)

[Implementation](<https://github.com/idanshen/Self-Distillation>)

Recorded evaluation configuration: {"Teacher Model":"Qwen2.5-3B-Instruct","Student Model":"Qwen2.5-0.5B"}. This record does not assert independent reproduction.

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