News Classification on AG News (test)
91.7AccuracyLabel-free
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Label-freeModel backbone=CLIP-large (395M)2026.03 | 91.7 | — | — | — | — | — | 1.426 | 20.7 | 4.7 | — | |
| Black-boxModel backbone=CLIP-large (395M)2026.03 | 91.5 | — | — | — | — | — | — | — | — | — | |
| CT-CBMModel backbone=CLIP-large (395M)2026.03 | 91.2 | — | — | — | — | — | 0.146 | 26.2 | 7.9 | — | |
| Label-freeModel backbone=CLIP-base (110M)2026.03 | 91 | — | — | — | — | — | 1.264 | 21.2 | 5 | — | |
| CT-CBMModel backbone=CLIP-base (110M)2026.03 | 91 | — | — | — | — | — | 0.101 | 24.4 | 5.9 | — | |
| SFTBackbone=Llama3.1-8B2026.05 | 91 | — | — | — | — | — | — | — | — | 69.7 | |
| f-CBMModel backbone=CLIP-large (395M)2026.03 | 90.7 | — | — | — | — | — | 0.068 | 0.4 | 0.2 | — | |
| Black-boxModel backbone=CLIP-base (110M)2026.03 | 90.6 | — | — | — | — | — | — | — | — | — | |
| f-CBMModel backbone=CLIP-base (110M)2026.03 | 90.6 | — | — | — | — | — | 0.056 | 0.5 | 0.6 | — | |
| SBRBackbone=Llama3.1-8B2026.05 | 90.1 | — | — | — | — | — | — | — | — | 5.1 | |
| BoosterBackbone=Llama3.1-8B2026.05 | 89.8 | — | — | — | — | — | — | — | — | 63.7 | |
| LisaBackbone=Llama3.1-8B2026.05 | 89.6 | — | — | — | — | — | — | — | — | 58.7 | |
| DeepAlignBackbone=Llama3.1-8B2026.05 | 89.4 | — | — | — | — | — | — | — | — | 30.1 | |
| VaccineBackbone=Llama3.1-8B2026.05 | 89.3 | — | — | — | — | — | — | — | — | 61.5 | |
| CONDACCBackbone=GPTJ-6B2022.12 | 87.3 | — | — | — | 2.6 | 70.5 | — | — | — | — | |
| UN-CONDACCBackbone=GPTJ-6B2022.12 | 87.1 | — | — | — | 1.1 | 84.6 | — | — | — | — | |
| CONDACCBackbone=OPT-13B2022.12 | 87 | — | — | — | 1 | 83.6 | — | — | — | — | |
| DATAMODELSBackbone=GPTJ-6B2022.12 | 86.9 | — | — | — | 1.4 | 82.8 | — | — | — | — | |
| Indep.-CBMModel backbone=CLIP-base (110M)2026.03 | 86.6 | — | — | — | — | — | 0.043 | 2.8 | 0.5 | — | |
| Indep.-CBMModel backbone=CLIP-large (395M)2026.03 | 86.2 | — | — | — | — | — | 0.042 | 92.2 | 0.4 | — | |
| DATAMODELSBackbone=OPT-13B2022.12 | 85.7 | — | — | — | 1.7 | 81.8 | — | — | — | — | |
| TOPPROMPTS-10Backbone=GPTJ-6B2022.12 | 85.4 | — | — | — | 2.4 | 76.3 | — | — | — | — | |
| ALL + CALIBBackbone=GPTJ-6B2022.12 | 85.2 | — | — | — | 2.7 | 78 | — | — | — | — | |
| UN-CONDACCBackbone=OPT-13B2022.12 | 84.6 | — | — | — | 3.1 | 72.5 | — | — | — | — | |
| TOPPROMPTS-5Backbone=GPTJ-6B2022.12 | 84.4 | — | — | — | 3.3 | 74.3 | — | — | — | — | |
| TOPPROMPTS-10Backbone=OPT-13B2022.12 | 84.4 | — | — | — | 3.5 | 70.9 | — | — | — | — | |
| ONESHOTBackbone=OPT-13B2022.12 | 84.2 | — | — | — | 2.9 | 73.1 | — | — | — | — | |
| ParBaLS-MAP EPIGinitial_labels=100, query_schedule=10 x 102025.10 | 84.11 | — | — | — | — | — | — | — | — | — | |
| TOPPROMPTS-5Backbone=OPT-13B2022.12 | 83.9 | — | — | — | 3.1 | 74 | — | — | — | — | |
| Confidenceinitial_labels=100, query_schedule=10 x 102025.10 | 83.78 | — | — | — | — | — | — | — | — | — | |
| GALAXYinitial_labels=100, query_schedule=10 x 102025.10 | 83.59 | — | — | — | — | — | — | — | — | — | |
| ParBaLS EPIGinitial_labels=100, query_schedule=10 x 102025.10 | 83.53 | — | — | — | — | — | — | — | — | — | |
| ALLBackbone=GPTJ-6B2022.12 | 83.5 | — | — | — | 3.8 | 70.4 | — | — | — | — | |
| ONESHOTBackbone=GPTJ-6B2022.12 | 83.3 | — | — | — | 3.4 | 71.9 | — | — | — | — | |
| RANDOMBackbone=GPTJ-6B2022.12 | 82.5 | — | — | — | 4.7 | 67.1 | — | — | — | — | |
| UN-ONESHOTBackbone=OPT-13B2022.12 | 82.3 | — | — | — | 4.5 | 64.9 | — | — | — | — | |
| ALL + CALIBBackbone=OPT-13B2022.12 | 81.8 | — | — | — | 4.1 | 70.7 | — | — | — | — | |
| UN-TOPPROMPTS-5Backbone=OPT-13B2022.12 | 81.8 | — | — | — | 6.9 | 51.8 | — | — | — | — | |
| ALLBackbone=OPT-13B2022.12 | 81.6 | — | — | — | 5.9 | 64.2 | — | — | — | — | |
| Randominitial_labels=100, query_schedule=10 x 102025.10 | 80.64 | — | — | — | — | — | — | — | — | — | |
| BADGEinitial_labels=100, query_schedule=10 x 102025.10 | 79.44 | — | — | — | — | — | — | — | — | — | |
| UN-ONESHOTBackbone=GPTJ-6B2022.12 | 78.1 | — | — | — | 4.2 | 69.8 | — | — | — | — | |
| RANDOMBackbone=OPT-13B2022.12 | 78 | — | — | — | 7.5 | 61.4 | — | — | — | — | |
| GLISTERinitial_labels=100, query_schedule=10 x 102025.10 | 77.92 | — | — | — | — | — | — | — | — | — | |
| UN-TOPPROMPTS-5Backbone=GPTJ-6B2022.12 | 76.4 | — | — | — | 9.8 | 53 | — | — | — | — | |
| UN-ALLBackbone=GPTJ-6B2022.12 | 75.1 | — | — | — | 9.9 | 46.5 | — | — | — | — | |
| CoreSetinitial_labels=100, query_schedule=10 x 102025.10 | 73.23 | — | — | — | — | — | — | — | — | — | |
| UN-ALLBackbone=OPT-13B2022.12 | 67.3 | — | — | — | 17.2 | 26.4 | — | — | — | — |