HyperAI

Document Summarization On Arxiv

Metrics

ROUGE-1

Results

Performance results of various models on this benchmark

Comparison Table
Model NameROUGE-1
sparsifying-transformer-models-with-trainable47.15