Results 1 to 10 of about 467,421 (344)
Fast Inference from Transformers via Speculative Decoding [PDF]
International Conference on Machine Learning, 2022 Inference from large autoregressive models like Transformers is slow - decoding K tokens takes K serial runs of the model. In this work we introduce speculative decoding - an algorithm to sample from autoregressive models faster without any changes to ...
Yaniv Leviathan +2 more
semanticscholar +1 more source
Accelerating Large Language Model Decoding with Speculative Sampling [PDF]
arXiv.org, 2023 We present speculative sampling, an algorithm for accelerating transformer decoding by enabling the generation of multiple tokens from each transformer call.
Charlie Chen +5 more
semanticscholar +1 more source
Contrastive Decoding: Open-ended Text Generation as Optimization [PDF]
Annual Meeting of the Association for Computational Linguistics, 2022 Given a language model (LM), maximum probability is a poor decoding objective for open-ended generation, because it produces short and repetitive text. On the other hand, sampling can often produce incoherent text that drifts from the original topics. We
Xiang Lisa Li +7 more
semanticscholar +1 more source
RNAi Degrades the SARS-CoV-2 Spike Protein RNA for Developing Drugs to Treat COVID-19
Nano Biomedicine and Engineering, 2022 COVID-19 is caused by severe acute respiratory SARS-CoV-2. Regardless of the availability of treatment strategies for COVID-19, effective therapy will remain essential.
Weiwei Zhang +17 more
doaj +1 more source
Scientific Reports, 2022 In the quest for more effective radiation treatment options that can improve both cell killing and healthy tissue recovery, combined radiation therapies are lately in the spotlight.
Pilar López-Nieva +8 more
doaj +1 more source
PICARD: Parsing Incrementally for Constrained Auto-Regressive Decoding from Language Models [PDF]
Conference on Empirical Methods in Natural Language Processing, 2021 Large pre-trained language models for textual data have an unconstrained output space; at each decoding step, they can produce any of 10,000s of sub-word tokens. When fine-tuned to target constrained formal languages like SQL, these models often generate
Torsten Scholak +2 more
semanticscholar +1 more source
Patterns of Differentially Expressed circRNAs in Human Thymocytes
Non-Coding RNA, 2022 Circular RNAs (circRNAs) are suggested to play a discriminative role between some stages of thymocyte differentiation. However, differential aspects of the stage of mature single-positive thymocytes remain to be explored.
Pilar López-Nieva +7 more
doaj +1 more source
DExperts: Decoding-Time Controlled Text Generation with Experts and Anti-Experts [PDF]
Annual Meeting of the Association for Computational Linguistics, 2021 Despite recent advances in natural language generation, it remains challenging to control attributes of generated text. We propose DExperts: Decoding-time Experts, a decoding-time method for controlled text generation that combines a pretrained language ...
Alisa Liu +6 more
semanticscholar +1 more source
Quantifying telomeric lncRNAs using PNA-labelled RNA-Flow FISH (RNA-Flow)
Communications Biology, 2022 Telomeric-associated long non-coding RNAs TERRA and TERC can be quantified in culture cell lines and human PBMCs using peptide nucleic acid probes with flow cytometry (RNA-Flow).
Iria González-Vasconcellos +4 more
doaj +1 more source
Generalized Decoding for Pixel, Image, and Language [PDF]
Computer Vision and Pattern Recognition, 2022 We present X-Decoder, a generalized decoding model that can predict pixel-level segmentation and language tokens seamlessly. X-Decoder takes as input two types of queries: (i) generic non-semantic queries and (ii) semantic queries induced from text ...
Xueyan Zou +13 more
semanticscholar +1 more source