ZeroGen: Efficient Zero-shot Learning via Dataset Generation
Open paper
Benchmark
Model
Rank
Results
data-free-knowledge-distillation-on-qnli
ZeroGen (T5-base)
#1
Accuracy: 88.5
data-free-knowledge-distillation-on-squad
ZeroGen (T5-base)
#2
Exact Match: 69.4
Rank counts only results with a code link.