Sam Thomson
YOU?
Author Swipe
View article: MICE for CATs: Model-Internal Confidence Estimation for Calibrating Agents with Tools
MICE for CATs: Model-Internal Confidence Estimation for Calibrating Agents with Tools Open
Tool-using agents that act in the world need to be both useful and safe. Well-calibrated model confidences can be used to weigh the risk versus reward of potential actions, but prior work shows that many models are poorly calibrated. Inspi…
View article: Toward Interactive Dictation
Toward Interactive Dictation Open
Voice dictation is an increasingly important text input modality. Existing systems that allow both dictation and editing-by-voice restrict their command language to flat templates invoked by trigger words. In this work, we study the feasib…
View article: Toward Interactive Dictation
Toward Interactive Dictation Open
Voice dictation is an increasingly important text input modality. Existing systems that allow both dictation and editing-by-voice restrict their command language to flat templates invoked by trigger words. In this work, we study the feasib…
View article: Ardagh Community Trust: transgressing boundaries, asserting community
Ardagh Community Trust: transgressing boundaries, asserting community Open
Administrative boundaries are ubiquitous. A vital technology of power within the modern nation-state’s mode of bureaucratic governance, they carve up and abstract land and water alike into conceptual totalities that, in their simplificatio…
View article: From a documented past of the Jersey breed in Africa to a profit index linked future
From a documented past of the Jersey breed in Africa to a profit index linked future Open
The paper reports on the prevalence and performance of the Jersey cattle breed in Africa, highlighting its geographic distribution and describing the reported performance and other related characteristics from the early 1900s to the presen…
View article: BenchCLAMP: A Benchmark for Evaluating Language Models on Syntactic and Semantic Parsing
BenchCLAMP: A Benchmark for Evaluating Language Models on Syntactic and Semantic Parsing Open
Recent work has shown that generation from a prompted or fine-tuned language model can perform well at semantic parsing when the output is constrained to be a valid semantic representation. We introduce BenchCLAMP, a Benchmark to evaluate …
View article: When More Data Hurts: A Troubling Quirk in Developing Broad-Coverage Natural Language Understanding Systems
When More Data Hurts: A Troubling Quirk in Developing Broad-Coverage Natural Language Understanding Systems Open
In natural language understanding (NLU) production systems, users' evolving needs necessitate the addition of new features over time, indexed by new symbols added to the meaning representation space. This requires additional training data …
View article: Online Semantic Parsing for Latency Reduction in Task-Oriented Dialogue
Online Semantic Parsing for Latency Reduction in Task-Oriented Dialogue Open
Jiawei Zhou, Jason Eisner, Michael Newman, Emmanouil Antonios Platanios, Sam Thomson. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2022.
View article: When More Data Hurts: A Troubling Quirk in Developing Broad-Coverage Natural Language Understanding Systems
When More Data Hurts: A Troubling Quirk in Developing Broad-Coverage Natural Language Understanding Systems Open
Elias Stengel-Eskin, Emmanouil Antonios Platanios, Adam Pauls, Sam Thomson, Hao Fang, Benjamin Van Durme, Jason Eisner, Yu Su. Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing. 2022.
View article: Guided K-best Selection for Semantic Parsing Annotation
Guided K-best Selection for Semantic Parsing Annotation Open
Anton Belyy, Chieh-yang Huang, Jacob Andreas, Emmanouil Antonios Platanios, Sam Thomson, Richard Shin, Subhro Roy, Aleksandr Nisnevich, Charles Chen, Benjamin Van Durme. Proceedings of the 60th Annual Meeting of the Association for Computa…
View article: Constrained Language Models Yield Few-Shot Semantic Parsers
Constrained Language Models Yield Few-Shot Semantic Parsers Open
We explore the use of large pretrained language models as few-shot semantic parsers. The goal in semantic parsing is to generate a structured meaning representation given a natural language input. However, language models are trained to ge…
View article: Value-Agnostic Conversational Semantic Parsing
Value-Agnostic Conversational Semantic Parsing Open
Emmanouil Antonios Platanios, Adam Pauls, Subhro Roy, Yuchen Zhang, Alexander Kyte, Alan Guo, Sam Thomson, Jayant Krishnamurthy, Jason Wolfe, Jacob Andreas, Dan Klein. Proceedings of the 59th Annual Meeting of the Association for Computati…
View article: Constrained Language Models Yield Few-Shot Semantic Parsers
Constrained Language Models Yield Few-Shot Semantic Parsers Open
Richard Shin, Christopher Lin, Sam Thomson, Charles Chen, Subhro Roy, Emmanouil Antonios Platanios, Adam Pauls, Dan Klein, Jason Eisner, Benjamin Van Durme. Proceedings of the 2021 Conference on Empirical Methods in Natural Language Proces…
View article: Compositional Generalization for Neural Semantic Parsing via Span-level Supervised Attention
Compositional Generalization for Neural Semantic Parsing via Span-level Supervised Attention Open
Pengcheng Yin, Hao Fang, Graham Neubig, Adam Pauls, Emmanouil Antonios Platanios, Yu Su, Sam Thomson, Jacob Andreas. Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human L…
View article: Syntactic Scaffolds for Semantic Structures
Syntactic Scaffolds for Semantic Structures Open
We introduce the syntactic scaffold, an approach to incorporating syntactic information into semantic tasks. Syntactic scaffolds avoid expensive syntactic processing at runtime, only making use of a treebank during training, through a mult…
View article: Rational Recurrences
Rational Recurrences Open
Despite the tremendous empirical success of neural models in natural language processing, many of them lack the strong intuitions that accompany classical machine learning approaches. Recently, connections have been shown between convoluti…
View article: Neural Motifs: Scene Graph Parsing with Global Context
Neural Motifs: Scene Graph Parsing with Global Context Open
We investigate the problem of producing structured graph representations of visual scenes. Our work analyzes the role of motifs: regularly appearing substructures in scene graphs. We present new quantitative insights on such repeated struc…
View article: SoPa: Bridging CNNs, RNNs, and Weighted Finite-State Machines
SoPa: Bridging CNNs, RNNs, and Weighted Finite-State Machines Open
Recurrent and convolutional neural networks comprise two distinct families of models that have proven to be useful for encoding natural language utterances. In this paper we present SoPa, a new model that aims to bridge these two approache…
View article: Backpropagating through Structured Argmax using a SPIGOT
Backpropagating through Structured Argmax using a SPIGOT Open
We introduce the structured projection of intermediate gradients optimization technique (SPIGOT), a new method for backpropagating through neural networks that include hard-decision structured predictions (e.g., parsing) in intermediate la…
View article: Learning Joint Semantic Parsers from Disjoint Data
Learning Joint Semantic Parsers from Disjoint Data Open
We present a new approach to learning semantic parsers from multiple datasets, even when the target semantic formalisms are drastically different, and the underlying corpora do not overlap. We handle such "disjoint" data by treating annota…
View article: Syntactic Scaffolds for Semantic Structures
Syntactic Scaffolds for Semantic Structures Open
We introduce the syntactic scaffold, an approach to incorporating syntactic information into semantic tasks. Syntactic scaffolds avoid expensive syntactic processing at runtime, only making use of a treebank during training, through a mult…
View article: Backpropagating through Structured Argmax using a SPIGOT
Backpropagating through Structured Argmax using a SPIGOT Open
We introduce structured projection of intermediate gradients (SPIGOT), a new method for backpropagating through neural networks that include hard-decision structured predictions (e.g., parsing) in intermediate layers. SPIGOT requires no ma…
View article: Bridging CNNs, RNNs, and Weighted Finite-State Machines
Bridging CNNs, RNNs, and Weighted Finite-State Machines Open
Recurrent and convolutional neural networks comprise two distinct families of models that have proven to be useful for encoding natural language utterances. In this paper we present SoPa, a new model that aims to bridge these two approache…
View article: Rational Recurrences
Rational Recurrences Open
Despite the tremendous empirical success of neural models in natural language processing, many of them lack the strong intuitions that accompany classical machine learning approaches. Recently, connections have been shown between convoluti…
View article: Learning Joint Semantic Parsers from Disjoint Data
Learning Joint Semantic Parsers from Disjoint Data Open
Hao Peng, Sam Thomson, Swabha Swayamdipta, Noah A. Smith. Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers). 2018.
View article: Frame-Semantic Parsing with Softmax-Margin Segmental RNNs and a Syntactic Scaffold
Frame-Semantic Parsing with Softmax-Margin Segmental RNNs and a Syntactic Scaffold Open
We present a new, efficient frame-semantic parser that labels semantic arguments to FrameNet predicates. Built using an extension to the segmental RNN that emphasizes recall, our basic system achieves competitive performance without any ca…