Vega v2 6B (KD-based prompt transfer)

Reported on 4 benchmarks across 3 tasks · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Natural Language Processing4 results

Question AnsweringonCOPA
Accuracy· 2022-12-04
99.4
best: 100 (PaLM 540B (finetuned) )
Toward Efficient Language Model Pretraining and Downstream Adaptation via Self-Evolution: A Case Study on SuperGLUE arXiv:2212.01853
Natural Language InferenceonCommitmentBank
Accuracy· 2022-12-04
99.2
best: 100 (PaLM 540B (finetuned))
Toward Efficient Language Model Pretraining and Downstream Adaptation via Self-Evolution: A Case Study on SuperGLUE arXiv:2212.01853
Natural Language InferenceonCommitmentBank
F1· 2022-12-04
98.6
best: 100 (PaLM 540B (finetuned))
Toward Efficient Language Model Pretraining and Downstream Adaptation via Self-Evolution: A Case Study on SuperGLUE arXiv:2212.01853
Coreference ResolutiononWinograd Schema Challenge
Accuracy· 2022-12-04
98.6
best: 100 (PaLM 540B (fine-tuned))
Toward Efficient Language Model Pretraining and Downstream Adaptation via Self-Evolution: A Case Study on SuperGLUE arXiv:2212.01853