# Natural Language Inference on ANLI Round 3 (Macro F1)

Source: [Natural Language Inference on ANLI Round 3 (Macro F1)](<https://www.sota2.com/research/sota/natural-language-inference-on-anli-round-3-macro-f1>)

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

Data updated: 2026-02-26T05:51:26.318Z

Evaluation of Natural Language Inference performance on the ANLI Round 3 dataset, specifically measured using the Macro F1 score.

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 | Natural Language Inference |
| Dataset (including any stated split) | ANLI Round 3 (R3) |
| Metric | Macro F1 |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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

- [Macro F1](<https://www.sota2.com/research/sota/natural-language-inference-on-anli-round-3-macro-f1.md?metric=Macro%20F1>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Macro F1 | Result date | Report paper |
| --- | --- | --- | --- |
| WCT - CV | 44.43 | Not recorded | 68fa73c21e209131cea5e4ed |
| CO-TEACHING | 43.83 | Not recorded | 68fa73c21e209131cea5e4ed |
| SIMPLE FT - ENSEMBLED | 43.37 | Not recorded | 68fa73c21e209131cea5e4ed |
| 15% CORRUPTED | 42.34 | Not recorded | 68fa73c21e209131cea5e4ed |

### WCT - CV

Macro F1: 44.43. Result ID: 6993218e3a6c31484913ed69.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73c21e209131cea5e4ed.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73c21e209131cea5e4ed.md>)

[Implementation](<https://github.com/msadat3/weighted_cotraining>)

Recorded evaluation configuration: {"Backbone":"RoBERTa","Label Corruption Rate":"15%"}. This record does not assert independent reproduction.

### CO-TEACHING

Macro F1: 43.83. Result ID: 6993218e3a6c31484913ed68.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73c21e209131cea5e4ed.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64fe1e209131cea07aa4.md>)

[Implementation](<https://github.com/bhanML/Co-teaching>)

Recorded evaluation configuration: {"Backbone":"RoBERTa","Label Corruption Rate":"15%"}. This record does not assert independent reproduction.

### SIMPLE FT - ENSEMBLED

Macro F1: 43.37. Result ID: 6993218e3a6c31484913ed67.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73c21e209131cea5e4ed.md>)

Recorded evaluation configuration: {"Backbone":"RoBERTa","Label Corruption Rate":"15%"}. This record does not assert independent reproduction.

### 15% CORRUPTED

Macro F1: 42.34. Result ID: 6993218e3a6c31484913ed66.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73c21e209131cea5e4ed.md>)

Recorded evaluation configuration: {"Backbone":"RoBERTa","Label Corruption Rate":"15%"}. This record does not assert independent reproduction.

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