# Compositional Zero-Shot Learning on MIT-States Closed World (test)

Source: [Compositional Zero-Shot Learning on MIT-States Closed World (test)](<https://www.sota2.com/research/sota/compositional-zero-shot-learning-on-mit-states-closed-world-test>)

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

Data updated: 2026-03-18T09:08:15.847Z

Evaluation of Compositional Zero-Shot Learning performance on the test split of the MIT-States Closed World dataset, reporting AUC, Harmonic Mean (HM), Score S, and Score U.

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 | Compositional Zero-Shot Learning |
| Dataset (including any stated split) | MIT-States Closed World (test) |
| Metric | AUC |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 27 of 27 |
| 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

- [AUC](<https://www.sota2.com/research/sota/compositional-zero-shot-learning-on-mit-states-closed-world-test.md?metric=AUC>): Higher is better; unit: PERCENT.

- [Harmonic Mean (HM)](<https://www.sota2.com/research/sota/compositional-zero-shot-learning-on-mit-states-closed-world-test.md?metric=Harmonic%20Mean%20(HM)>): Higher is better; unit: PERCENT.

- [Score S](<https://www.sota2.com/research/sota/compositional-zero-shot-learning-on-mit-states-closed-world-test.md?metric=Score%20S>): Higher is better; unit: PERCENT.

- [Score U](<https://www.sota2.com/research/sota/compositional-zero-shot-learning-on-mit-states-closed-world-test.md?metric=Score%20U>): Higher is better; unit: PERCENT.

- [Seen Score](<https://www.sota2.com/research/sota/compositional-zero-shot-learning-on-mit-states-closed-world-test.md?metric=Seen%20Score>): Higher is better; unit: PERCENT.

- [Unseen Score](<https://www.sota2.com/research/sota/compositional-zero-shot-learning-on-mit-states-closed-world-test.md?metric=Unseen%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | AUC | Result date | Report paper |
| --- | --- | --- | --- |
| Troika + FlowComposer | 23.5 | Not recorded | 69ba31920636e5bf4afc08b7 |
| CAILA | 23.4 | Not recorded | 68fa711d1e209131cea4f2d4 |
| PLO | 23.4 | Not recorded | 69ba31920636e5bf4afc08b7 |
| LOGICZSL | 23.4 | Not recorded | 69ba31920636e5bf4afc08b7 |
| CDS-CZSL | 22.4 | Not recorded | 69ba31920636e5bf4afc08b7 |
| PLID | 22.1 | Not recorded | 69ba31920636e5bf4afc08b7 |
| Troika | 22.1 | Not recorded | 69ba31920636e5bf4afc08b7 |
| IMAX | 21.9 | Not recorded | 69ba31920636e5bf4afc08b7 |
| DFSP(BiF) | 20.8 | Not recorded | 69ba31920636e5bf4afc08b7 |
| DFSP(i2t) | 20.7 | Not recorded | 69ba31920636e5bf4afc08b7 |
| CSP + FlowComposer | 20.7 | Not recorded | 69ba31920636e5bf4afc08b7 |
| DFSP | 20.6 | Not recorded | 68fa711d1e209131cea4f2d4 |
| DFSP(t2i) | 20.6 | Not recorded | 69ba31920636e5bf4afc08b7 |
| DLM | 20 | Not recorded | 69ba31920636e5bf4afc08b7 |
| CSP | 19.6 | Not recorded | 69ba31920636e5bf4afc08b7 |
| CSP | 19.4 | Not recorded | 68fa711d1e209131cea4f2d4 |
| Co-CGE | 17 | Not recorded | 68fa711d1e209131cea4f2d4 |
| CoOp | 13.5 | Not recorded | 68fa711d1e209131cea4f2d4 |
| CoOP | 13.5 | Not recorded | 69ba31920636e5bf4afc08b7 |
| CLIP-ZS | 11 | Not recorded | 68fa711d1e209131cea4f2d4 |
| CLIP | 11 | Not recorded | 69ba31920636e5bf4afc08b7 |
| CAPE | 6.7 | Not recorded | 68fa711d1e209131cea4f2d4 |
| Co-CGE | 6.6 | Not recorded | 68fa711d1e209131cea4f2d4 |
| CGE | 6.5 | Not recorded | 68fa711d1e209131cea4f2d4 |
| OADis | 5.9 | Not recorded | 68fa711d1e209131cea4f2d4 |
| SCEN | 5.3 | Not recorded | 68fa711d1e209131cea4f2d4 |
| CompCos | 4.5 | Not recorded | 68fa711d1e209131cea4f2d4 |

### Troika + FlowComposer

AUC: 23.5. Result ID: 69ba6b6147f34ce924a48aa8.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

Recorded evaluation configuration: {"baseline":"Troika"}. This record does not assert independent reproduction.

### CAILA

AUC: 23.4. Result ID: 69a1590535dfcabb00e672f4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa711d1e209131cea4f2d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa711d1e209131cea4f2d4.md>)

[Implementation](<https://www.sota2.com/github.com/zhaohengz/CAILA>)

Recorded evaluation configuration: {"CLIP-based":"true","Backbone":"ViT-L/14"}. This record does not assert independent reproduction.

### PLO

AUC: 23.4. Result ID: 69ba6b6147f34ce924a48aa3.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7aae1e209131cea84922.md>)

Recorded evaluation configuration: {"leveraging LLM knowledge":"true"}. This record does not assert independent reproduction.

### LOGICZSL

AUC: 23.4. Result ID: 69ba6b6147f34ce924a48aa4.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

Recorded evaluation configuration: {"leveraging LLM knowledge":"true"}. This record does not assert independent reproduction.

### CDS-CZSL

AUC: 22.4. Result ID: 69ba6b6147f34ce924a48aa0.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72521e209131cea559d4.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### PLID

AUC: 22.1. Result ID: 69ba6b6147f34ce924a48aa2.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74101e209131cea5f8f2.md>)

Recorded evaluation configuration: {"leveraging LLM knowledge":"true"}. This record does not assert independent reproduction.

### Troika

AUC: 22.1. Result ID: 69ba6b6147f34ce924a48aa7.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72ad1e209131cea57c90.md>)

Recorded evaluation configuration: {"implementation source":"re-implemented"}. This record does not assert independent reproduction.

### IMAX

AUC: 21.9. Result ID: 69ba6b6147f34ce924a48aa1.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### DFSP(BiF)

AUC: 20.8. Result ID: 69ba6b6147f34ce924a48a9d.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6df81e209131cea3d0c9.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### DFSP(i2t)

AUC: 20.7. Result ID: 69ba6b6147f34ce924a48a9c.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6df81e209131cea3d0c9.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### CSP + FlowComposer

AUC: 20.7. Result ID: 69ba6b6147f34ce924a48aa6.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

Recorded evaluation configuration: {"baseline":"CSP"}. This record does not assert independent reproduction.

### DFSP

AUC: 20.6. Result ID: 69a1590535dfcabb00e672f3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa711d1e209131cea4f2d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6df81e209131cea3d0c9.md>)

Recorded evaluation configuration: {"CLIP-based":"true","Backbone":"ViT-L/14"}. This record does not assert independent reproduction.

### DFSP(t2i)

AUC: 20.6. Result ID: 69ba6b6147f34ce924a48a9e.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6df81e209131cea3d0c9.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### DLM

AUC: 20. Result ID: 69ba6b6147f34ce924a48a9f.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### CSP

AUC: 19.6. Result ID: 69ba6b6147f34ce924a48aa5.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f291e209131cea44272.md>)

Recorded evaluation configuration: {"implementation source":"re-implemented"}. This record does not assert independent reproduction.

### CSP

AUC: 19.4. Result ID: 69a1590535dfcabb00e672f2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa711d1e209131cea4f2d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f291e209131cea44272.md>)

Recorded evaluation configuration: {"CLIP-based":"true","Backbone":"ViT-L/14"}. This record does not assert independent reproduction.

### Co-CGE

AUC: 17. Result ID: 69a1590535dfcabb00e672f1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa711d1e209131cea4f2d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c371e209131cea32d6e.md>)

Recorded evaluation configuration: {"CLIP-based":"true","Backbone":"ViT-L/14"}. This record does not assert independent reproduction.

### CoOp

AUC: 13.5. Result ID: 69a1590535dfcabb00e672f0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa711d1e209131cea4f2d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d8b1e209131cea3a814.md>)

[Implementation](<https://github.com/KaiyangZhou/CoOp>)

Recorded evaluation configuration: {"CLIP-based":"true","Backbone":"ViT-L/14"}. This record does not assert independent reproduction.

### CoOP

AUC: 13.5. Result ID: 69ba6b6147f34ce924a48a9b.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d8b1e209131cea3a814.md>)

[Implementation](<https://github.com/KaiyangZhou/CoOp>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### CLIP-ZS

AUC: 11. Result ID: 69a1590535dfcabb00e672ef.

[Reporting paper](<https://www.sota2.com/research/paper/68fa711d1e209131cea4f2d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69581e209131cea21ad0.md>)

[Implementation](<https://github.com/OpenAI/CLIP>)

Recorded evaluation configuration: {"CLIP-based":"true","Backbone":"ViT-L/14"}. This record does not assert independent reproduction.

### CLIP

AUC: 11. Result ID: 69ba6b6147f34ce924a48a9a.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31920636e5bf4afc08b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69581e209131cea21ad0.md>)

[Implementation](<https://github.com/OpenAI/CLIP>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### CAPE

AUC: 6.7. Result ID: 69a1590535dfcabb00e672ee.

[Reporting paper](<https://www.sota2.com/research/paper/68fa711d1e209131cea4f2d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6dbf1e209131cea3bd01.md>)

Recorded evaluation configuration: {"CLIP-based":"false"}. This record does not assert independent reproduction.

### Co-CGE

AUC: 6.6. Result ID: 69a1590535dfcabb00e672ed.

[Reporting paper](<https://www.sota2.com/research/paper/68fa711d1e209131cea4f2d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c371e209131cea32d6e.md>)

Recorded evaluation configuration: {"CLIP-based":"false"}. This record does not assert independent reproduction.

### CGE

AUC: 6.5. Result ID: 69a1590535dfcabb00e672ec.

[Reporting paper](<https://www.sota2.com/research/paper/68fa711d1e209131cea4f2d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69e91e209131cea24d40.md>)

[Implementation](<https://github.com/ExplainableML/czsl>)

Recorded evaluation configuration: {"CLIP-based":"false"}. This record does not assert independent reproduction.

### OADis

AUC: 5.9. Result ID: 69a1590535dfcabb00e672ea.

[Reporting paper](<https://www.sota2.com/research/paper/68fa711d1e209131cea4f2d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c9c1e209131cea34e1e.md>)

[Implementation](<https://github.com/nirat1606/OADis>)

Recorded evaluation configuration: {"CLIP-based":"false"}. This record does not assert independent reproduction.

### SCEN

AUC: 5.3. Result ID: 69a1590535dfcabb00e672eb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa711d1e209131cea4f2d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ce61e209131cea36a24.md>)

[Implementation](<https://github.com/XDUxyLi/SCEN-master>)

Recorded evaluation configuration: {"CLIP-based":"false"}. This record does not assert independent reproduction.

### CompCos

AUC: 4.5. Result ID: 69a1590535dfcabb00e672e9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa711d1e209131cea4f2d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69991e209131cea22f2b.md>)

Recorded evaluation configuration: {"CLIP-based":"false"}. This record does not assert independent reproduction.

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