Unlocking Compositional Generalization in Pre-trained Models Using Intermediate Representations

Benchmark Model Rank Results
semantic-parsing-on-cfqT5-3B w/ Intermediate Representations#2Exact Match: 83.8