ProGen: Progressive Zero-shot Dataset Generation via In-context Feedback
Open paper
Benchmark
Model
Rank
Results
data-free-knowledge-distillation-on-qnli
ProGen (T5-base)
#2
Accuracy: 85.9
data-free-knowledge-distillation-on-squad
ProGen (T5-base)
#3
Exact Match: 68.1
Rank counts only results with a code link.