Exact sentence-level MAP decoding is NP-hard and exact conditioned normalization is #P-hard for general autoregressive models under global constraints, unlike local sampling or finite-state Markov models.
On NMT Search Errors and Model Errors: Cat Got Your Tongue? InProceedings of EMNLP-IJCNLP 2019, pages 3356–3362, 2019
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Hidden Biases in Conditioning Autoregressive Models
Exact sentence-level MAP decoding is NP-hard and exact conditioned normalization is #P-hard for general autoregressive models under global constraints, unlike local sampling or finite-state Markov models.