Search (76 results, page 2 of 4)

0.003754243 = product of:
  0.011262729 = sum of:
    0.011262729 = weight(_text_:a in 255) [ClassicSimilarity], result of:
      0.011262729 = score(doc=255,freq=16.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.2161963 = fieldWeight in 255, product of:
          4.0 = tf(freq=16.0), with freq of:
            16.0 = termFreq=16.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.046875 = fieldNorm(doc=255)
  0.33333334 = coord(1/3)

Abstract

A basic distinction in agent-based data-driven Database Semantics (DBS) is between language and nonlanguage cognition. Language cognition transfers content between agents by means of raw data. Nonlanguage cognition maps between content and raw data inside the focus agent. {\it Recognition} applies a concept type to raw data, resulting in a concept token. In language recognition, the focus agent (hearer) takes raw language-data (surfaces) produced by another agent (speaker) as input, while nonlanguage recognition takes raw nonlanguage-data as input. In either case, the output is a content which is stored in the agent's onboard short term memory. {\it Action} adapts a concept type to a purpose, resulting in a token. In language action, the focus agent (speaker) produces language-dependent surfaces for another agent (hearer), while nonlanguage action produces intentions for a nonlanguage purpose. In either case, the output is raw action data. As long as the procedural implementation of place holder values works properly, it is compatible with the DBS requirement of input-output equivalence between the natural prototype and the artificial reconstruction.

0.003754243 = product of:
  0.011262729 = sum of:
    0.011262729 = weight(_text_:a in 853) [ClassicSimilarity], result of:
      0.011262729 = score(doc=853,freq=4.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.2161963 = fieldWeight in 853, product of:
          2.0 = tf(freq=4.0), with freq of:
            4.0 = termFreq=4.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.09375 = fieldNorm(doc=853)
  0.33333334 = coord(1/3)

Abstract

A new chatbot from OpenAI is inspiring awe, fear, stunts and attempts to circumvent its guardrails.

Type

a

0.003754243 = product of:
  0.011262729 = sum of:
    0.011262729 = weight(_text_:a in 871) [ClassicSimilarity], result of:
      0.011262729 = score(doc=871,freq=16.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.2161963 = fieldWeight in 871, product of:
          4.0 = tf(freq=16.0), with freq of:
            16.0 = termFreq=16.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.046875 = fieldNorm(doc=871)
  0.33333334 = coord(1/3)

Abstract

Natural language processing tasks, such as question answering, machine translation, reading comprehension, and summarization, are typically approached with supervised learning on task-specific datasets. We demonstrate that language models begin to learn these tasks without any explicit supervision when trained on a new dataset of millions of webpages called WebText. When conditioned on a document plus questions, the answers generated by the language model reach 55 F1 on the CoQA dataset - matching or exceeding the performance of 3 out of 4 baseline systems without using the 127,000+ training examples. The capacity of the language model is essential to the success of zero-shot task transfer and increasing it improves performance in a log-linear fashion across tasks. Our largest model, GPT-2, is a 1.5B parameter Transformer that achieves state of the art results on 7 out of 8 tested language modeling datasets in a zero-shot setting but still underfits WebText. Samples from the model reflect these improvements and contain coherent paragraphs of text. These findings suggest a promising path towards building language processing systems which learn to perform tasks from their naturally occurring demonstrations.

Type

a

0.0036484683 = product of:
  0.010945405 = sum of:
    0.010945405 = weight(_text_:a in 2804) [ClassicSimilarity], result of:
      0.010945405 = score(doc=2804,freq=34.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.21010503 = fieldWeight in 2804, product of:
          5.8309517 = tf(freq=34.0), with freq of:
            34.0 = termFreq=34.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.03125 = fieldNorm(doc=2804)
  0.33333334 = coord(1/3)

Abstract

The ACL Anthology is a digital archive of conference and journal papers in natural language processing and computational linguistics. Its primary purpose is to serve as a reference repository of research results, but we believe that it can also be an object of study and a platform for research in its own right. We describe an enriched and standardized reference corpus derived from the ACL Anthology that can be used for research in scholarly document processing. This corpus, which we call the ACL Anthology Reference Corpus (ACL ARC), brings together the recent activities of a number of research groups around the world. Our goal is to make the corpus widely available, and to encourage other researchers to use it as a standard testbed for experiments in both bibliographic and bibliometric research.

Content

Vgl. auch: Automatic Term Recognition (ATR) is a research task that deals with the identification of domain-specific terms. Terms, in simple words, are textual realization of significant concepts in an expertise domain. Additionally, domain-specific terms may be classified into a number of categories, in which each category represents a significant concept. A term classification task is often defined on top of an ATR procedure to perform such categorization. For instance, in the biomedical domain, terms can be classified as drugs, proteins, and genes. This is a reference dataset for terminology extraction and classification research in computational linguistics. It is a set of manually annotated terms in English language that are extracted from the ACL Anthology Reference Corpus (ACL ARC). The ACL ARC is a canonicalised and frozen subset of scientific publications in the domain of Human Language Technologies (HLT). It consists of 10,921 articles from 1965 to 2006. The dataset, called ACL RD-TEC, is comprised of more than 69,000 candidate terms that are manually annotated as valid and invalid terms. Furthermore, valid terms are classified as technology and non-technology terms. Technology terms refer to a method, process, or in general a technological concept in the domain of HLT, e.g. machine translation, word sense disambiguation, and language modelling. On the other hand, non-technology terms refer to important concepts other than technological; examples of such terms in the domain of HLT are multilingual lexicon, corpora, word sense, and language model. The dataset is created to serve as a gold standard for the comparison of the algorithms of term recognition and classification. [http://catalog.elra.info/product_info.php?products_id=1236].

Type

a

0.003462655 = product of:
  0.010387965 = sum of:
    0.010387965 = weight(_text_:a in 1150) [ClassicSimilarity], result of:
      0.010387965 = score(doc=1150,freq=10.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.19940455 = fieldWeight in 1150, product of:
          3.1622777 = tf(freq=10.0), with freq of:
            10.0 = termFreq=10.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.0546875 = fieldNorm(doc=1150)
  0.33333334 = coord(1/3)

Abstract

Voice search is becoming a popular mode for interacting with search engines. As a result, research has gone into building better voice transcription engines, interfaces, and search engines that better handle inherent verbosity of queries. However, when one considers its use by non- native speakers of English, another aspect that becomes important is the formulation of the query by users. In this paper, we present the results of a preliminary study that we conducted with non-native English speakers who formulate queries for given retrieval tasks. Our results show that the current search engines are sensitive in their rankings to the query formulation, and thus highlights the need for developing more robust ranking methods.

Type

a

0.00325127 = product of:
  0.009753809 = sum of:
    0.009753809 = weight(_text_:a in 1061) [ClassicSimilarity], result of:
      0.009753809 = score(doc=1061,freq=12.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.18723148 = fieldWeight in 1061, product of:
          3.4641016 = tf(freq=12.0), with freq of:
            12.0 = termFreq=12.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.046875 = fieldNorm(doc=1061)
  0.33333334 = coord(1/3)

Abstract

The object of Information Retrieval is to retrieve all relevant documents for a user query and only those relevant documents. Much research has focused on achieving this objective with little regard for storage overhead or performance. In the paper we evaluate the use of Part of Speech Tagging to improve, the index storage overhead and general speed of the system with only a minimal reduction to precision recall measurements. We tagged 500Mbs of the Los Angeles Times 1990 and 1989 document collection provided by TREC for parts of speech. We then experimented to find the most relevant part of speech to index. We show that 90% of precision recall is achieved with 40% of the document collections terms. We also show that this is a improvement in overhead with only a 1% reduction in precision recall.

Type

a

0.00325127 = product of:
  0.009753809 = sum of:
    0.009753809 = weight(_text_:a in 2697) [ClassicSimilarity], result of:
      0.009753809 = score(doc=2697,freq=12.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.18723148 = fieldWeight in 2697, product of:
          3.4641016 = tf(freq=12.0), with freq of:
            12.0 = termFreq=12.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.046875 = fieldNorm(doc=2697)
  0.33333334 = coord(1/3)

Abstract

Text mining and natural language processing are fast growing areas of research, with numerous applications in business, science and creative industries. This paper presents TextFlows, a web-based text mining and natural language processing platform supporting workflow construction, sharing and execution. The platform enables visual construction of text mining workflows through a web browser, and the execution of the constructed workflows on a processing cloud. This makes TextFlows an adaptable infrastructure for the construction and sharing of text processing workflows, which can be reused in various applications. The paper presents the implemented text mining and language processing modules, and describes some precomposed workflows. Their features are demonstrated on three use cases: comparison of document classifiers and of different part-of-speech taggers on a text categorization problem, and outlier detection in document corpora.

Type

a

0.00325127 = product of:
  0.009753809 = sum of:
    0.009753809 = weight(_text_:a in 870) [ClassicSimilarity], result of:
      0.009753809 = score(doc=870,freq=12.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.18723148 = fieldWeight in 870, product of:
          3.4641016 = tf(freq=12.0), with freq of:
            12.0 = termFreq=12.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.046875 = fieldNorm(doc=870)
  0.33333334 = coord(1/3)

Abstract

Natural language understanding comprises a wide range of diverse tasks such as textual entailment, question answering, semantic similarity assessment, and document classification. Although large unlabeled text corpora are abundant, labeled data for learning these specific tasks is scarce, making it challenging for discriminatively trained models to perform adequately. We demonstrate that large gains on these tasks can be realized by generative pre-training of a language model on a diverse corpus of unlabeled text, followed by discriminative fine-tuning on each specific task. In contrast to previous approaches, we make use of task-aware input transformations during fine-tuning to achieve effective transfer while requiring minimal changes to the model architecture. We demonstrate the effectiveness of our approach on a wide range of benchmarks for natural language understanding. Our general task-agnostic model outperforms discriminatively trained models that use architectures specifically crafted for each task, significantly improving upon the state of the art in 9 out of the 12 tasks studied. For instance, we achieve absolute improvements of 8.9% on commonsense reasoning (Stories Cloze Test), 5.7% on question answering (RACE), and 1.5% on textual entailment (MultiNLI).

Type

a

0.00325127 = product of:
  0.009753809 = sum of:
    0.009753809 = weight(_text_:a in 873) [ClassicSimilarity], result of:
      0.009753809 = score(doc=873,freq=12.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.18723148 = fieldWeight in 873, product of:
          3.4641016 = tf(freq=12.0), with freq of:
            12.0 = termFreq=12.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.046875 = fieldNorm(doc=873)
  0.33333334 = coord(1/3)

Abstract

In this review paper, ChatGPT, a public tool developed by OpenAI that utilizes GPT technology to fulfill a range of text-based requests is examined. ChatGPT is a sophisticated chatbot capable of understanding and interpreting user requests, generating appropriate responses in nearly natural human language, and completing advanced tasks such as writing thank you letters and addressing productivity issues. The details of how ChatGPT works, as well as the potential impacts of this technology on various industries, are discussed. The concept of Generative Pre-Trained Transformer (GPT), the language model on which ChatGPT is based, is also explored, as well as the process of unsupervised pretraining and supervised fine-tuning that is used to refine the GPT algorithm. A letter written by ChatGPT to a colleague from Iran is presented as an example of the chatbot's capabilities.

0.0030970925 = product of:
  0.009291277 = sum of:
    0.009291277 = weight(_text_:a in 773) [ClassicSimilarity], result of:
      0.009291277 = score(doc=773,freq=8.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.17835285 = fieldWeight in 773, product of:
          2.828427 = tf(freq=8.0), with freq of:
            8.0 = termFreq=8.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.0546875 = fieldNorm(doc=773)
  0.33333334 = coord(1/3)

Abstract

We show that generating English Wikipedia articles can be approached as a multi-document summarization of source documents. We use extractive summarization to coarsely identify salient information and a neural abstractive model to generate the article. For the abstractive model, we introduce a decoder-only architecture that can scalably attend to very long sequences, much longer than typical encoder- decoder architectures used in sequence transduction. We show that this model can generate fluent, coherent multi-sentence paragraphs and even whole Wikipedia articles. When given reference documents, we show it can extract relevant factual information as reflected in perplexity, ROUGE scores and human evaluations.

Type

a

0.003065327 = product of:
  0.009195981 = sum of:
    0.009195981 = weight(_text_:a in 3671) [ClassicSimilarity], result of:
      0.009195981 = score(doc=3671,freq=6.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.17652355 = fieldWeight in 3671, product of:
          2.4494898 = tf(freq=6.0), with freq of:
            6.0 = termFreq=6.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.0625 = fieldNorm(doc=3671)
  0.33333334 = coord(1/3)

Abstract

Semantic networks produced from human data have statistical properties that cannot be easily captured by spatial representations. We explore a probabilistic approach to semantic representation that explicitly models the probability with which words occurin diffrent contexts, and hence captures the probabilistic relationships between words. We show that this representation has statistical properties consistent with the large-scale structure of semantic networks constructed by humans, and trace the origins of these properties.

Type

a

0.003065327 = product of:
  0.009195981 = sum of:
    0.009195981 = weight(_text_:a in 3365) [ClassicSimilarity], result of:
      0.009195981 = score(doc=3365,freq=6.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.17652355 = fieldWeight in 3365, product of:
          2.4494898 = tf(freq=6.0), with freq of:
            6.0 = termFreq=6.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.0625 = fieldNorm(doc=3365)
  0.33333334 = coord(1/3)

Abstract

We present a novel fine-grained semantic representation of text and an approach to constructing it. This representation is largely extractable by today's technologies and facilitates more detailed semantic analysis. We discuss the requirements driving the representation, suggest how it might be of value in the automated tutoring domain, and provide evidence of its validity.

Type

a

0.003065327 = product of:
  0.009195981 = sum of:
    0.009195981 = weight(_text_:a in 872) [ClassicSimilarity], result of:
      0.009195981 = score(doc=872,freq=24.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.17652355 = fieldWeight in 872, product of:
          4.8989797 = tf(freq=24.0), with freq of:
            24.0 = termFreq=24.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.03125 = fieldNorm(doc=872)
  0.33333334 = coord(1/3)

Abstract

Recent work has demonstrated substantial gains on many NLP tasks and benchmarks by pre-training on a large corpus of text followed by fine-tuning on a specific task. While typically task-agnostic in architecture, this method still requires task-specific fine-tuning datasets of thousands or tens of thousands of examples. By contrast, humans can generally perform a new language task from only a few examples or from simple instructions - something which current NLP systems still largely struggle to do. Here we show that scaling up language models greatly improves task-agnostic, few-shot performance, sometimes even reaching competitiveness with prior state-of-the-art fine-tuning approaches. Specifically, we train GPT-3, an autoregressive language model with 175 billion parameters, 10x more than any previous non-sparse language model, and test its performance in the few-shot setting. For all tasks, GPT-3 is applied without any gradient updates or fine-tuning, with tasks and few-shot demonstrations specified purely via text interaction with the model. GPT-3 achieves strong performance on many NLP datasets, including translation, question-answering, and cloze tasks, as well as several tasks that require on-the-fly reasoning or domain adaptation, such as unscrambling words, using a novel word in a sentence, or performing 3-digit arithmetic. At the same time, we also identify some datasets where GPT-3's few-shot learning still struggles, as well as some datasets where GPT-3 faces methodological issues related to training on large web corpora. Finally, we find that GPT-3 can generate samples of news articles which human evaluators have difficulty distinguishing from articles written by humans. We discuss broader societal impacts of this finding and of GPT-3 in general.

Type

a

0.00296799 = product of:
  0.00890397 = sum of:
    0.00890397 = weight(_text_:a in 3094) [ClassicSimilarity], result of:
      0.00890397 = score(doc=3094,freq=10.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.1709182 = fieldWeight in 3094, product of:
          3.1622777 = tf(freq=10.0), with freq of:
            10.0 = termFreq=10.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.046875 = fieldNorm(doc=3094)
  0.33333334 = coord(1/3)

Abstract

The paper addresses the issue of how online natural language question answering, based on deep semantic analysis, may compete with currently popular keyword search, open domain information retrieval systems, covering a horizontal domain. We suggest the multiagent question answering approach, where each domain is represented by an agent which tries to answer questions taking into account its specific knowledge. The meta-agent controls the cooperation between question answering agents and chooses the most relevant answer(s). We argue that multiagent question answering is optimal in terms of access to business and financial knowledge, flexibility in query phrasing, and efficiency and usability of advice. The knowledge and advice encoded in the system are initially prepared by domain experts. We analyze the commercial application of multiagent question answering and the robustness of the meta-agent. The paper suggests that a multiagent architecture is optimal when a real world question answering domain combines a number of vertical ones to form a horizontal domain.

0.00296799 = product of:
  0.00890397 = sum of:
    0.00890397 = weight(_text_:a in 2803) [ClassicSimilarity], result of:
      0.00890397 = score(doc=2803,freq=10.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.1709182 = fieldWeight in 2803, product of:
          3.1622777 = tf(freq=10.0), with freq of:
            10.0 = termFreq=10.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.046875 = fieldNorm(doc=2803)
  0.33333334 = coord(1/3)

Abstract

This paper introduces ACL RD-TEC: a dataset for evaluating the extraction and classification of terms from literature in the domain of computational linguistics. The dataset is derived from the Association for Computational Linguistics anthology reference corpus (ACL ARC). In its first release, the ACL RD-TEC consists of automatically segmented, part-of-speech-tagged ACL ARC documents, three lists of candidate terms, and more than 82,000 manually annotated terms. The annotated terms are marked as either valid or invalid, and valid terms are further classified as technology and non-technology terms. Technology terms signify methods, algorithms, and solutions in computational linguistics. The paper describes the dataset and reports the relevant statistics. We hope the step described in this paper encourages a collaborative effort towards building a full-fledged annotated corpus from the computational linguistics literature.

Type

a

0.00296799 = product of:
  0.00890397 = sum of:
    0.00890397 = weight(_text_:a in 970) [ClassicSimilarity], result of:
      0.00890397 = score(doc=970,freq=10.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.1709182 = fieldWeight in 970, product of:
          3.1622777 = tf(freq=10.0), with freq of:
            10.0 = termFreq=10.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.046875 = fieldNorm(doc=970)
  0.33333334 = coord(1/3)

Abstract

The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task, improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature. We show that the Transformer generalizes well to other tasks by applying it successfully to English constituency parsing both with large and limited training data.

Type

a

0.0027982478 = product of:
  0.008394743 = sum of:
    0.008394743 = weight(_text_:a in 851) [ClassicSimilarity], result of:
      0.008394743 = score(doc=851,freq=20.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.16114321 = fieldWeight in 851, product of:
          4.472136 = tf(freq=20.0), with freq of:
            20.0 = termFreq=20.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.03125 = fieldNorm(doc=851)
  0.33333334 = coord(1/3)

Abstract

Literature review articles are essential to summarize the related work in the selected field. However, covering all related studies takes too much time and effort. This study questions how Artificial Intelligence can be used in this process. We used ChatGPT to create a literature review article to show the stage of the OpenAI ChatGPT artificial intelligence application. As the subject, the applications of Digital Twin in the health field were chosen. Abstracts of the last three years (2020, 2021 and 2022) papers were obtained from the keyword "Digital twin in healthcare" search results on Google Scholar and paraphrased by ChatGPT. Later on, we asked ChatGPT questions. The results are promising; however, the paraphrased parts had significant matches when checked with the Ithenticate tool. This article is the first attempt to show the compilation and expression of knowledge will be accelerated with the help of artificial intelligence. We are still at the beginning of such advances. The future academic publishing process will require less human effort, which in turn will allow academics to focus on their studies. In future studies, we will monitor citations to this study to evaluate the academic validity of the content produced by the ChatGPT. 1. Introduction OpenAI ChatGPT (ChatGPT, 2022) is a chatbot based on the OpenAI GPT-3 language model. It is designed to generate human-like text responses to user input in a conversational context. OpenAI ChatGPT is trained on a large dataset of human conversations and can be used to create responses to a wide range of topics and prompts. The chatbot can be used for customer service, content creation, and language translation tasks, creating replies in multiple languages. OpenAI ChatGPT is available through the OpenAI API, which allows developers to access and integrate the chatbot into their applications and systems. OpenAI ChatGPT is a variant of the GPT (Generative Pre-trained Transformer) language model developed by OpenAI. It is designed to generate human-like text, allowing it to engage in conversation with users naturally and intuitively. OpenAI ChatGPT is trained on a large dataset of human conversations, allowing it to understand and respond to a wide range of topics and contexts. It can be used in various applications, such as chatbots, customer service agents, and language translation systems. OpenAI ChatGPT is a state-of-the-art language model able to generate coherent and natural text that can be indistinguishable from text written by a human. As an artificial intelligence, ChatGPT may need help to change academic writing practices. However, it can provide information and guidance on ways to improve people's academic writing skills.

0.0027916813 = product of:
  0.008375044 = sum of:
    0.008375044 = weight(_text_:a in 923) [ClassicSimilarity], result of:
      0.008375044 = score(doc=923,freq=26.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.16076508 = fieldWeight in 923, product of:
          5.0990195 = tf(freq=26.0), with freq of:
            26.0 = termFreq=26.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.02734375 = fieldNorm(doc=923)
  0.33333334 = coord(1/3)

Abstract

"I still don't know what to think about GPT-4, the new large language model (LLM) from OpenAI. On the one hand it is a remarkable product that easily passes the Turing test. If you ask it questions, via the ChatGPT interface, GPT-4 can easily produce fluid sentences largely indistinguishable from those a person might write. But on the other hand, amid the exceptional levels of hype and anticipation, it's hard to know where GPT-4 and other LLMs truly fit in the larger project of making machines intelligent.
They might appear intelligent, but LLMs are nothing of the sort. They don't understand the meanings of the words they are using, nor the concepts expressed within the sentences they create. When asked how to bring a cow back to life, earlier versions of ChatGPT, for example, which ran on a souped-up version of GPT-3, would confidently provide a list of instructions. So-called hallucinations like this happen because language models have no concept of what a "cow" is or that "death" is a non-reversible state of being. LLMs do not have minds that can think about objects in the world and how they relate to each other. All they "know" is how likely it is that some sets of words will follow other sets of words, having calculated those probabilities from their training data. To make sense of all this, I spoke with Gary Marcus, an emeritus professor of psychology and neural science at New York University, for "Babbage", our science and technology podcast. Last year, as the world was transfixed by the sudden appearance of ChatGPT, he made some fascinating predictions about GPT-4.
He doesn't dismiss the potential of LLMs to become useful assistants in all sorts of ways-Google and Microsoft have already announced that they will be integrating LLMs into their search and office productivity software. But he talked me through some of his criticisms of the technology's apparent capabilities. At the heart of Dr Marcus's thoughtful critique is an attempt to put LLMs into proper context. Deep learning, the underlying technology that makes LLMs work, is only one piece of the puzzle in the quest for machine intelligence. To reach the level of artificial general intelligence (AGI) that many tech companies strive for-i.e. machines that can plan, reason and solve problems in the way human brains can-they will need to deploy a suite of other AI techniques. These include, for example, the kind of "symbolic AI" that was popular before artificial neural networks and deep learning became all the rage.
People use symbols to think about the world: if I say the words "cat", "house" or "aeroplane", you know instantly what I mean. Symbols can also be used to describe the way things are behaving (running, falling, flying) or they can represent how things should behave in relation to each other (a "+" means add the numbers before and after). Symbolic AI is a way to embed this human knowledge and reasoning into computer systems. Though the idea has been around for decades, it fell by the wayside a few years ago as deep learning-buoyed by the sudden easy availability of lots of training data and cheap computing power-became more fashionable. In the near future at least, there's no doubt people will find LLMs useful. But whether they represent a critical step on the path towards AGI, or rather just an intriguing detour, remains to be seen."

0.0027093915 = product of:
  0.008128175 = sum of:
    0.008128175 = weight(_text_:a in 972) [ClassicSimilarity], result of:
      0.008128175 = score(doc=972,freq=12.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.15602624 = fieldWeight in 972, product of:
          3.4641016 = tf(freq=12.0), with freq of:
            12.0 = termFreq=12.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.0390625 = fieldNorm(doc=972)
  0.33333334 = coord(1/3)

Abstract

Believable proxies of human behavior can empower interactive applications ranging from immersive environments to rehearsal spaces for interpersonal communication to prototyping tools. In this paper, we introduce generative agents--computational software agents that simulate believable human behavior. Generative agents wake up, cook breakfast, and head to work; artists paint, while authors write; they form opinions, notice each other, and initiate conversations; they remember and reflect on days past as they plan the next day. To enable generative agents, we describe an architecture that extends a large language model to store a complete record of the agent's experiences using natural language, synthesize those memories over time into higher-level reflections, and retrieve them dynamically to plan behavior. We instantiate generative agents to populate an interactive sandbox environment inspired by The Sims, where end users can interact with a small town of twenty five agents using natural language. In an evaluation, these generative agents produce believable individual and emergent social behaviors: for example, starting with only a single user-specified notion that one agent wants to throw a Valentine's Day party, the agents autonomously spread invitations to the party over the next two days, make new acquaintances, ask each other out on dates to the party, and coordinate to show up for the party together at the right time. We demonstrate through ablation that the components of our agent architecture--observation, planning, and reflection--each contribute critically to the believability of agent behavior. By fusing large language models with computational, interactive agents, this work introduces architectural and interaction patterns for enabling believable simulations of human behavior.

Type

a

0.002682161 = product of:
  0.008046483 = sum of:
    0.008046483 = weight(_text_:a in 156) [ClassicSimilarity], result of:
      0.008046483 = score(doc=156,freq=6.0), product of:
        0.05209492 = queryWeight, product of:
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.045180224 = queryNorm
        0.1544581 = fieldWeight in 156, product of:
          2.4494898 = tf(freq=6.0), with freq of:
            6.0 = termFreq=6.0
          1.153047 = idf(docFreq=37942, maxDocs=44218)
          0.0546875 = fieldNorm(doc=156)
  0.33333334 = coord(1/3)

Abstract

In this work, we present the results of research for evaluating a methodology for extracting mathematical terms for precalculus using the techniques for semantically-oriented statistical search. We use the corpus-based approach and the combination of different statistically-based techniques for extracting keywords, collocations and co-occurrences incorporated in the Sketch Engine software. We evaluate the collocations candidate terms for the basic concept function(s) and approve the related methodology by precalculus domain conceptual terms definitions. Finally, we offer a conceptual terms hierarchical representation and discuss the results with respect to their possible applications.

Type

a