Hierarchical Policy Optimization post-trains LLMs for simultaneous speech translation on imperfect data, yielding over +7 COMET and +1.25 MetricX improvements at 1.5-second latency on English-to-Chinese/German/Japanese tasks.
Can we start generating the program and executing it before the user even finishes the utterance so that the faster response can be achieved by the system?
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Hierarchical Policy Optimization for Simultaneous Translation of Unbounded Speech
Hierarchical Policy Optimization post-trains LLMs for simultaneous speech translation on imperfect data, yielding over +7 COMET and +1.25 MetricX improvements at 1.5-second latency on English-to-Chinese/German/Japanese tasks.