paper
-with-
me
Papers
Browse State-of-the-Art
Datasets
Methods
AI Agents
Trends
Digest
🌙
Cross-Lingual Transfer
벤치마크
Cross-Lingual Transfer on
XCOPA
12개 결과 ·
⬇ CSV
·
JSON
Accuracy
55.5
65.22
74.95
84.68
94.4
2020-04
2026-09
MAD-X Base — 60.94 (2020-04-30)
MAD-X Base — 60.94 (2020-04-30)
RoBERTa Large (translate test) — 76.05 (2020-05-01)
RoBERTa Large (translate test) — 76.05 (2020-05-01)
mGPT — 55.5 (2022-04-15)
mGPT — 55.5 (2022-04-15)
mT0-13B — 84.45 (2022-11-03)
BLOOMZ — 75.5 (2022-11-03)
mT0-13B — 84.45 (2022-11-03)
BLOOMZ — 75.5 (2022-11-03)
PaLM 2 (few-shot) — 94.4 (2023-05-17)
PaLM 2 (few-shot) — 94.4 (2023-05-17)
MAD-X Base — 60.94 (2020-04-30)
RoBERTa Large (translate test) — 76.05 (2020-05-01)
mT0-13B — 84.45 (2022-11-03)
PaLM 2 (few-shot) — 94.4 (2023-05-17)
2020-04-30 — MAD-X Base: Accuracy 60.94
2020-05-01 — RoBERTa Large (translate test): Accuracy 76.05
2022-11-03 — mT0-13B: Accuracy 84.45
2023-05-17 — PaLM 2 (few-shot): Accuracy 94.4
Rank
Model
Accuracy
Extra Training Data
Paper
Code
Year
1
PaLM 2 (few-shot)
94.4
PaLM 2 Technical Report
eternityyw/tram-benchmark
2023
2
mT0-13B
84.45
Crosslingual Generalization through Multitask Finetuning
bigscience-workshop/xmtf
2022
3
RoBERTa Large (translate test)
76.05
✓
XCOPA: A Multilingual Dataset for Causal Commonsense Reasoning
cambridgeltl/xcopa
2020
4
BLOOMZ
75.5
Crosslingual Generalization through Multitask Finetuning
bigscience-workshop/xmtf
2022
5
MAD-X Base
60.94
✓
MAD-X: An Adapter-Based Framework for Multi-Task Cross-Lingual Transfer
Adapter-Hub/adapter-transformers
·
cambridgeltl/xcopa
·
aaronsom/wmt21-qe-tudarmstadt
2020
6
mGPT
55.5
mGPT: Few-Shot Learners Go Multilingual
ai-forever/mgpt
2022
7
PaLM 2 (few-shot)
94.4
PaLM 2 Technical Report
eternityyw/tram-benchmark
2023
8
mT0-13B
84.45
Crosslingual Generalization through Multitask Finetuning
bigscience-workshop/xmtf
2022
9
RoBERTa Large (translate test)
76.05
✓
XCOPA: A Multilingual Dataset for Causal Commonsense Reasoning
cambridgeltl/xcopa
2020
10
BLOOMZ
75.5
Crosslingual Generalization through Multitask Finetuning
bigscience-workshop/xmtf
2022
11
MAD-X Base
60.94
✓
MAD-X: An Adapter-Based Framework for Multi-Task Cross-Lingual Transfer
Adapter-Hub/adapter-transformers
·
cambridgeltl/xcopa
·
aaronsom/wmt21-qe-tudarmstadt
2020
12
mGPT
55.5
mGPT: Few-Shot Learners Go Multilingual
ai-forever/mgpt
2022
1–12 / 12
페이지당
10
20
50
100