Generative Pretrained Structured Transformers: Unsupervised Syntactic Language Models at Scale
Open paper
Benchmark
Model
Rank
Results
constituency-grammar-induction-on-ptb
GPST(left to right parsing)
#16
Mean F1 (WSJ): 55.2
natural-language-inference-on-multinli
GPST(unsupervised generative syntactic LM)
#36
Matched: 81.8
Mismatched: 82.0
Rank counts only results with a code link.