Value-based Search in Execution Space for Mapping Instructions to Programs
| Autor: | Jonathan Herzig, Dor Muhlgay, Jonathan Berant |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
FOS: Computer and information sciences
Computer Science - Computation and Language Theoretical computer science Computer science Value (computer science) 02 engineering and technology 010501 environmental sciences Space (commercial competition) 01 natural sciences Search algorithm 0202 electrical engineering electronic engineering information engineering Beam search 020201 artificial intelligence & image processing State (computer science) Computation and Language (cs.CL) Natural language 0105 earth and related environmental sciences |
| Zdroj: | NAACL-HLT (1) |
| DOI: | 10.18653/v1/n19-1193 |
| Popis: | Training models to map natural language instructions to programs, given target world supervision only, requires searching for good programs at training time. Search is commonly done using beam search in the space of partial programs or program trees, but as the length of the instructions grows finding a good program becomes difficult. In this work, we propose a search algorithm that uses the target world state, known at training time, to train a critic network that predicts the expected reward of every search state. We then score search states on the beam by interpolating their expected reward with the likelihood of programs represented by the search state. Moreover, we search not in the space of programs but in a more compressed state of program executions, augmented with recent entities and actions. On the SCONE dataset, we show that our algorithm dramatically improves performance on all three domains compared to standard beam search and other baselines. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|