Zobrazeno 1 - 10
of 19 266
pro vyhledávání: '"Beam search"'
Beam search decoding is the de-facto method for decoding auto-regressive Neural Machine Translation (NMT) models, including multilingual NMT where the target language is specified as an input. However, decoding multilingual NMT models commonly produc
Externí odkaz:
http://arxiv.org/abs/2408.05738
Tokenising continuous speech into sequences of discrete tokens and modelling them with language models (LMs) has led to significant success in text-to-speech (TTS) synthesis. Although these models can generate speech with high quality and naturalness
Externí odkaz:
http://arxiv.org/abs/2408.16373
We adapt the well-known beam-search algorithm for machine translation to operate in a cascaded real-time speech translation system. This proved to be more complex than initially anticipated, due to four key challenges: (1) real-time processing of int
Externí odkaz:
http://arxiv.org/abs/2407.11010
Autor:
Sudo, Yui, Shakeel, Muhammad, Fukumoto, Yosuke, Yan, Brian, Shi, Jiatong, Peng, Yifan, Watanabe, Shinji
End-to-end automatic speech recognition (E2E-ASR) can be classified into several network architectures, such as connectionist temporal classification (CTC), recurrent neural network transducer (RNN-T), attention-based encoder-decoder, and mask-predic
Externí odkaz:
http://arxiv.org/abs/2406.02950
Large Language Models (LLMs) have facilitated structured data generation, with applications in domains like tabular data, document databases, product catalogs, etc. However, concerns persist about generation veracity due to incorrect references or ha
Externí odkaz:
http://arxiv.org/abs/2406.00069
This paper addresses the problem of determining the maximum number of edges in a unit distance graph (UDG) of $n$ vertices using computer search. An unsolved problem of Paul Erd\H{o}s asks the maximum number of edges $u(n)$ a UDG of $n$ vertices can
Externí odkaz:
http://arxiv.org/abs/2406.15317
Large language models are revolutionizing several areas, including artificial creativity. However, the process of generation in machines profoundly diverges from that observed in humans. In particular, machine generation is characterized by a lack of
Externí odkaz:
http://arxiv.org/abs/2405.00099
Autor:
Pastukhov, Sergey
This paper introduces a novel algorithm for two-player deterministic games with perfect information, which we call PROBS (Predict Results of Beam Search). Unlike existing methods that predominantly rely on Monte Carlo Tree Search (MCTS) for decision
Externí odkaz:
http://arxiv.org/abs/2404.16072
Beam search with masked language models (MLMs) is challenging in part because joint probability distributions over sequences are not readily available, unlike for autoregressive models. However, estimating such distributions has important domain-spec
Externí odkaz:
http://arxiv.org/abs/2402.15020
Publikováno v:
Advances in Transportation Studies. SpecialIssue2024, p37-46. 10p.