Filters








1,860 Hits in 8.6 sec

Semantic features of object concepts generated with GPT-3 [article]

Hannes Hansen, Martin N. Hebart
2022 arXiv   pre-print
To this end, we probed a GPT-3 model to generate semantic features for 1,854 objects and compared automatically-generated features to existing human feature norms.  ...  GPT-3 generated many more features than humans, yet showed a similar distribution in the types of generated features.  ...  To this end, we used GPT-3 to generate a semantic feature norm for 1,854 diverse concepts of concrete objects. We chose concrete objects for two reasons.  ... 
arXiv:2202.03753v2 fatcat:q2x25gojavcfvjmabbnqpmc4ty

An ontology-based system for intelligent matching of travellers' needs for Group Package Tours

Dimitris N. Kanellopoulos
2008 International Journal of Digital Culture and Electronic Tourism  
The objective of this research work was to design an intelligent web portal to serve as service provider in tasks related with package tours.  ...  For this purpose, the knowledge of the package tour domain has been represented by means of ontology.  ...  Section 2 analyses critical service features in GPT provided by travel agencies. Section 3 presents some issues on ontological engineering and the use of ontologies in web portals.  ... 
doi:10.1504/ijdcet.2008.020136 fatcat:mcok7cfpjnef3msw73umrler2y

Word meaning in minds and machines [article]

Brenden M. Lake, Gregory L. Murphy
2021 arXiv   pre-print
We discuss more promising approaches to grounding NLP systems and argue that they will be more successful with a more human-like, conceptual basis for word meaning.  ...  In this article, we compare how humans and machines represent the meaning of words.  ...  Authors' Note Order of authorship is alphabetical.  ... 
arXiv:2008.01766v3 fatcat:vi4zp7ebxfcepesrz5b2vkxdcu

Deriving Contextualised Semantic Features from BERT (and Other Transformer Model) Embeddings [article]

Jacob Turton, David Vinson, Robert Elliott Smith
2020 arXiv   pre-print
Binder and colleagues proposed an intuitive embedding space where each dimension is based on one of 65 core semantic features.  ...  It additionally provides insights into how semantic features are represented across the different layers of the BERT model.  ...  Semantic feature values for the property words abrasive and babbling with different objects FEATURE PROPERTY OBJECT Method The property-object word pairs were fed into the transformer models as the  ... 
arXiv:2012.15353v1 fatcat:vmtvxxglcjgoppy4xpkcm6mq7y

Comparative Evaluation of Pretrained Transfer Learning Models on Automatic Short Answer Grading [article]

Sasi Kiran Gaddipati, Deebul Nair, Paul G. Plöger
2020 arXiv   pre-print
Previous works implemented the methods of concept mapping, facet mapping, and some used the conventional word embeddings for extracting semantic features.  ...  We train with a single feature, cosine similarity, extracted from the embeddings of these models.  ...  ACKNOWLEDGEMENTS We thank Tim Metzler for providing the updated version of the Mohler dataset for experimentation.  ... 
arXiv:2009.01303v1 fatcat:mqckyjsd2bhpfba34xmxg3ysbm

NLX-GPT: A Model for Natural Language Explanations in Vision and Vision-Language Tasks [article]

Fawaz Sammani, Tanmoy Mukherjee, Nikos Deligiannis
2022 arXiv   pre-print
We first conduct pre-training on large scale data of image-caption pairs for general understanding of images, and then formulate the answer as a text prediction task along with the explanation.  ...  We introduce NLX-GPT, a general, compact and faithful language model that can simultaneously predict an answer and explain it.  ...  bottom-up object level features [3] .  ... 
arXiv:2203.05081v1 fatcat:hh7zkumhlng35lgecvbgk6jbku

Natural Language Rationales with Full-Stack Visual Reasoning: From Pixels to Semantic Frames to Commonsense Graphs [article]

Ana Marasović, Chandra Bhagavatula, Jae Sung Park, Ronan Le Bras, Noah A. Smith, Yejin Choi
2020 arXiv   pre-print
We present Rationale^VT Transformer, an integrated model that learns to generate free-text rationales by combining pretrained language models with object recognition, grounded visual semantic frames, and  ...  The key challenge of accurate rationalization is comprehensive image understanding at all levels: not just their explicit content at the pixel level, but their contextual contents at the semantic and pragmatic  ...  Acknowledgments The authors thank Sarah Pratt for her assistance with the grounded situation recognizer, Amandalynne Paullada, members of the AllenNLP team, and anonymous reviewers for helpful feedback  ... 
arXiv:2010.07526v1 fatcat:6vafbtt34rccrfly327tjr4pwe

A drawback for substitutional arguments

Justina Diaz-Legaspe, Adam Sennet
2021 Language Sciences  
Competing theories on the semantics of group pejorative terms (also known as 'slurs') comprise both advocates and opponents to the Identity Thesis (IT), according to which these terms and their neutral  ...  counterparts do not differ in semantic value.  ...  With this narrow concept of semantics in play, the question on the meaning of GPTs can be stated more clearly: do slurring terms for groups have semantic values that involve negative properties characterizing  ... 
doi:10.1016/j.langsci.2021.101431 fatcat:n5gqynfvyrfdlogrtlvd3vttye

Controlling dynamic module composition through an extensible meta-level API

Eyvind W. Axelsen, Stein Krogdahl, Birger Møller-Pedersen
2010 SIGPLAN notices  
Dynamic languages have traditionally employed strong support for meta-programming, with hooks to control OO concepts such as method invocation and object construction, by utilizing meta-classes and meta-object  ...  We wish to support a wide range of possible composition semantics, and to make such choices available to the developer through a meta-level API.  ...  We will start with a general overview of the main concepts of the mechanism and its application and scope in Sections 3.1 through 3.3.  ... 
doi:10.1145/1899661.1869641 fatcat:al4xrykwqvbe7nqk6n5bz555da

Controlling dynamic module composition through an extensible meta-level API

Eyvind W. Axelsen, Stein Krogdahl, Birger Møller-Pedersen
2010 Proceedings of the 6th symposium on Dynamic languages - DLS '10  
Dynamic languages have traditionally employed strong support for meta-programming, with hooks to control OO concepts such as method invocation and object construction, by utilizing meta-classes and meta-object  ...  We wish to support a wide range of possible composition semantics, and to make such choices available to the developer through a meta-level API.  ...  We will start with a general overview of the main concepts of the mechanism and its application and scope in Sections 3.1 through 3.3.  ... 
doi:10.1145/1869631.1869641 dblp:conf/dls/AxelsenKM10 fatcat:4xnvluhlbbfo5fufjgnfeluhp4

Probing Multimodal Embeddings for Linguistic Properties: the Visual-Semantic Case [article]

Adam Dahlgren Lindström, Suna Bensch, Johanna Björklund, Frank Drewes
2021 arXiv   pre-print
To address this problem, we generalize the notion of probing tasks to the visual-semantic case.  ...  To this end, we (i) discuss the formalization of probing tasks for embeddings of image-caption pairs, (ii) define three concrete probing tasks within our general framework, (iii) train classifiers to probe  ...  We thank the anonymous reviewers for their valuable feedback, which has had a substantial influence on the final version of the paper.  ... 
arXiv:2102.11115v1 fatcat:f6l5gzk7hjdgjayvzgnw7ux4ha

Multimodal Story Generation on Plural Images [article]

Jing Jiang
2021 arXiv   pre-print
The output samples from the model demonstrate the ability to generate meaningful paragraphs of text containing the extracted features from the input images. This is an undergraduate project report.  ...  In this work, we propose the architecture to use images instead of text as the input of the text generation model, called StoryGen.  ...  GPT-2 is a direct scale-up of the previous GPT, with more than 10X the parameters and trained on more than 10X the amount of data.  ... 
arXiv:2001.10980v2 fatcat:qu3ygu3fgbhz7gnl3wv23v34gm

Language Models Can See: Plugging Visual Controls in Text Generation [article]

Yixuan Su and Tian Lan and Yahui Liu and Fangyu Liu and Dani Yogatama and Yan Wang and Lingpeng Kong and Nigel Collier
2022 arXiv   pre-print
Generative language models (LMs) such as GPT-2/3 can be prompted to generate text with remarkable quality.  ...  During decoding, MAGIC influences the generation of the LM by introducing a CLIP-induced score, called magic score, which regularizes the generated result to be semantically related to a given image while  ...  The evaluation follows a 5-point Likert scale (1, 2, 3, 4, or 5) for each of the following features: 15 • Coherence (coh.): Whether the generated story is semantically consistent with the title. • Fluency  ... 
arXiv:2205.02655v2 fatcat:f2q23myhgbhlpdqcujwg5aw2x4

Automated Bot Detection Based on Coherence Metric

Oleksandr Marchenko, Mariam Isoieva
2021 International Conference "Information Technology and Interactions"  
A classifier with a set of features based on a coherence metric and syntactic characteristics has been built. The method can be extended and used for different languages.  ...  One of the main distinguishing features of high-quality texts is coherence. This is also what automatically generated texts, especially long ones, often lack in.  ...  These models, especially GPT-3, allow one to generate texts that are very similar to those written by humans.  ... 
dblp:conf/iti2/MarchenkoI21 fatcat:uugwaccpwffjvppbklqgn3otta

ZeroCap: Zero-Shot Image-to-Text Generation for Visual-Semantic Arithmetic [article]

Yoad Tewel, Yoav Shalev, Idan Schwartz, Lior Wolf
2022 arXiv   pre-print
This is done by combining the visual-semantic model with a large language model, benefiting from the knowledge in both web-scale models.  ...  While such models can provide a powerful score for matching and subsequent zero-shot tasks, they are not capable of generating caption given an image.  ...  The contribution of the first author is part of a PhD thesis at Tel Aviv University.  ... 
arXiv:2111.14447v2 fatcat:vfowvq2qofbt3aef4l5swdanum
« Previous Showing results 1 — 15 out of 1,860 results