Natural Language Inference on XNLI Ur (dev)
70.6AccuracyXITE-reg-lda
Evaluation Results
| Method | Links | |
|---|---|---|
| XITE-reg-ldaBackbone=XLMR-large, Input Language=en2026.04 | 70.6 | |
| SkylineBackbone=XLMR-large, Input Language=en2026.04 | 68.11 | |
| Tpl-rand-kEvaluation protocol=translate-train, Supervision type=hard prompt labels, Data selection strategy=rand-k2024.07 | 66.43 | |
| Tpl-top-kEvaluation protocol=translate-train, Supervision type=hard prompt labels, Data selection strategy=top-k2024.07 | 66.27 | |
| Tpl-div-kEvaluation protocol=translate-train, Supervision type=hard prompt labels, Data selection strategy=div-k2024.07 | 65.91 | |
| T-div-kEvaluation protocol=translate-train, Supervision type=teacher soft pseudolabels, Data selection strategy=div-k2024.07 | 65.82 | |
| SRC+GENEvaluation protocol=translate-train2024.07 | 65.28 | |
| T-top-kEvaluation protocol=translate-train, Supervision type=teacher soft pseudolabels, Data selection strategy=top-k2024.07 | 64.4 | |
| SRC+GENZero-shot=true2024.07 | 63.92 | |
| Tpl-rand-kselection strategy=rand-k, Zero-shot=true2024.07 | 63.78 | |
| Tpl-top-kselection strategy=top-k, Zero-shot=true2024.07 | 63.33 | |
| T-rand-kEvaluation protocol=translate-train, Supervision type=teacher soft pseudolabels, Data selection strategy=rand-k2024.07 | 63.01 | |
| T-top-kselection strategy=top-k, Zero-shot=true2024.07 | 62.98 | |
| T-div-kselection strategy=div-k, Zero-shot=true2024.07 | 62.83 | |
| SkylineBackbone=XLMR-large, Input Language=ur2026.04 | 62.45 | |
| Tpl-div-kselection strategy=div-k, Zero-shot=true2024.07 | 62.33 | |
| SRCEvaluation protocol=translate-train2024.07 | 61.65 | |
| T-rand-kselection strategy=rand-k, Zero-shot=true2024.07 | 59.78 | |
| GENZero-shot=true2024.07 | 57.21 | |
| SRCZero-shot=true2024.07 | 57.15 | |
| GENEvaluation protocol=translate-train2024.07 | 56.61 | |
| XITE-reg-regBackbone=XLMR-large, Input Language=en2026.04 | 53.45 | |
| XITE-reg-ldaBackbone=XLMR-large, Input Language=ur2026.04 | 51.22 | |
| XITE-reg-regBackbone=XLMR-large, Input Language=ur2026.04 | 38.16 | |
| Baseline-PSBackbone=XLMR-large, Input Language=ur2026.04 | 35.1 | |
| Baseline-PSBackbone=XLMR-large, Input Language=en2026.04 | 34.5 |