Grounded SCAN poses a simple task, where an agent must execute action sequences based on a synthetic language instruction.
20 PAPERS • NO BENCHMARKS YET
A new English language dataset structured for task-oriented evaluation on unseen tasks.
3 PAPERS • NO BENCHMARKS YET
Official dataset of Decrypting Cryptic Crosswords: Semantically Complex Wordplay Puzzles as a Target for NLP.
1 PAPER • NO BENCHMARKS YET