publication_date
stringlengths 20
20
| doc_url
stringlengths 33
34
| id
stringlengths 33
34
| authors
stringlengths 5
20.1k
| update_date
stringlengths 20
20
| category_all
stringlengths 5
138
| abstract
stringlengths 123
3.09k
| category_primary
stringclasses 78
values | title
stringlengths 10
211
|
|---|---|---|---|---|---|---|---|---|
2023-02-13T18:00:44Z | http://arxiv.org/pdf/2302.07735v1 | http://arxiv.org/abs/2302.07735v1 | Ali Al-Kaswan, Maliheh Izadi, Arie van Deursen | 2023-02-13T18:00:44Z | cs.CL, cs.AI, cs.CR | Previous work has shown that Large Language Models are susceptible to
so-called data extraction attacks. This allows an attacker to extract a sample
that was contained in the training data, which has massive privacy
implications. The construction of data extraction attacks is challenging,
current attacks are quite inefficient, and there exists a significant gap in
the extraction capabilities of untargeted attacks and memorization. Thus,
targeted attacks are proposed, which identify if a given sample from the
training data, is extractable from a model. In this work, we apply a targeted
data extraction attack to the SATML2023 Language Model Training Data Extraction
Challenge. We apply a two-step approach. In the first step, we maximise the
recall of the model and are able to extract the suffix for 69% of the samples.
In the second step, we use a classifier-based Membership Inference Attack on
the generations. Our AutoSklearn classifier achieves a precision of 0.841. The
full approach reaches a score of 0.405 recall at a 10% false positive rate,
which is an improvement of 34% over the baseline of 0.301. | cs.CL | Targeted Attack on GPT-Neo for the SATML Language Model Data Extraction Challenge |
2023-02-13T18:00:23Z | http://arxiv.org/pdf/2302.06560v1 | http://arxiv.org/abs/2302.06560v1 | Yash Verma, Anubhav Jangra, Raghvendra Kumar, Sriparna Saha | 2023-02-13T18:00:23Z | cs.CL, cs.MM | Significant developments in techniques such as encoder-decoder models have
enabled us to represent information comprising multiple modalities. This
information can further enhance many downstream tasks in the field of
information retrieval and natural language processing; however, improvements in
multi-modal techniques and their performance evaluation require large-scale
multi-modal data which offers sufficient diversity. Multi-lingual modeling for
a variety of tasks like multi-modal summarization, text generation, and
translation leverages information derived from high-quality multi-lingual
annotated data. In this work, we present the current largest multi-lingual
multi-modal summarization dataset (M3LS), and it consists of over a million
instances of document-image pairs along with a professionally annotated
multi-modal summary for each pair. It is derived from news articles published
by British Broadcasting Corporation(BBC) over a decade and spans 20 languages,
targeting diversity across five language roots, it is also the largest
summarization dataset for 13 languages and consists of cross-lingual
summarization data for 2 languages. We formally define the multi-lingual
multi-modal summarization task utilizing our dataset and report baseline scores
from various state-of-the-art summarization techniques in a multi-lingual
setting. We also compare it with many similar datasets to analyze the
uniqueness and difficulty of M3LS. | cs.CL | Large Scale Multi-Lingual Multi-Modal Summarization Dataset |
2023-02-13T17:57:50Z | http://arxiv.org/pdf/2302.07267v6 | http://arxiv.org/abs/2302.07267v6 | Peter S. Park, Philipp Schoenegger, Chongyang Zhu | 2023-09-13T07:44:42Z | cs.HC, cs.AI, cs.CL, 68T50, I.2.7 | We test whether Large Language Models (LLMs) can be used to simulate human
participants in social-science studies. To do this, we run replications of 14
studies from the Many Labs 2 replication project with OpenAI's text-davinci-003
model, colloquially known as GPT3.5. Based on our pre-registered analyses, we
find that among the eight studies we could analyse, our GPT sample replicated
37.5% of the original results and 37.5% of the Many Labs 2 results. However, we
were unable to analyse the remaining six studies due to an unexpected
phenomenon we call the "correct answer" effect. Different runs of GPT3.5
answered nuanced questions probing political orientation, economic preference,
judgement, and moral philosophy with zero or near-zero variation in responses:
with the supposedly "correct answer." In one exploratory follow-up study, we
found that a "correct answer" was robust to changing the demographic details
that precede the prompt. In another, we found that most but not all "correct
answers" were robust to changing the order of answer choices. One of our most
striking findings occurred in our replication of the Moral Foundations Theory
survey results, where we found GPT3.5 identifying as a political conservative
in 99.6% of the cases, and as a liberal in 99.3% of the cases in the
reverse-order condition. However, both self-reported 'GPT conservatives' and
'GPT liberals' showed right-leaning moral foundations. Our results cast doubts
on the validity of using LLMs as a general replacement for human participants
in the social sciences. Our results also raise concerns that a hypothetical
AI-led future may be subject to a diminished diversity-of-thought. | cs.HC | Diminished Diversity-of-Thought in a Standard Large Language Model |
2023-02-13T17:55:54Z | http://arxiv.org/pdf/2302.06555v1 | http://arxiv.org/abs/2302.06555v1 | Jiaang Li, Yova Kementchedjhieva, Anders Søgaard | 2023-02-13T17:55:54Z | cs.CL | Large-scale pretrained language models (LMs) are said to ``lack the ability
to connect [their] utterances to the world'' (Bender and Koller, 2020). If so,
we would expect LM representations to be unrelated to representations in
computer vision models. To investigate this, we present an empirical evaluation
across three different LMs (BERT, GPT2, and OPT) and three computer vision
models (VMs, including ResNet, SegFormer, and MAE). Our experiments show that
LMs converge towards representations that are partially isomorphic to those of
VMs, with dispersion, and polysemy both factoring into the alignability of
vision and language spaces. We discuss the implications of this finding. | cs.CL | Implications of the Convergence of Language and Vision Model Geometries |
2023-02-13T17:34:13Z | http://arxiv.org/pdf/2302.06541v2 | http://arxiv.org/abs/2302.06541v2 | Maximilian Mozes, Jessica Hoffmann, Katrin Tomanek, Muhamed Kouate, Nithum Thain, Ann Yuan, Tolga Bolukbasi, Lucas Dixon | 2023-10-21T11:49:09Z | cs.CL | Text-based safety classifiers are widely used for content moderation and
increasingly to tune generative language model behavior - a topic of growing
concern for the safety of digital assistants and chatbots. However, different
policies require different classifiers, and safety policies themselves improve
from iteration and adaptation. This paper introduces and evaluates methods for
agile text classification, whereby classifiers are trained using small,
targeted datasets that can be quickly developed for a particular policy.
Experimenting with 7 datasets from three safety-related domains, comprising 15
annotation schemes, led to our key finding: prompt-tuning large language
models, like PaLM 62B, with a labeled dataset of as few as 80 examples can
achieve state-of-the-art performance. We argue that this enables a paradigm
shift for text classification, especially for models supporting safer online
discourse. Instead of collecting millions of examples to attempt to create
universal safety classifiers over months or years, classifiers could be tuned
using small datasets, created by individuals or small organizations, tailored
for specific use cases, and iterated on and adapted in the time-span of a day. | cs.CL | Towards Agile Text Classifiers for Everyone |
2023-02-13T16:11:58Z | http://arxiv.org/pdf/2302.06488v1 | http://arxiv.org/abs/2302.06488v1 | Yang Janet Liu, Amir Zeldes | 2023-02-13T16:11:58Z | cs.CL | Recent advances in discourse parsing performance create the impression that,
as in other NLP tasks, performance for high-resource languages such as English
is finally becoming reliable. In this paper we demonstrate that this is not the
case, and thoroughly investigate the impact of data diversity on RST parsing
stability. We show that state-of-the-art architectures trained on the standard
English newswire benchmark do not generalize well, even within the news domain.
Using the two largest RST corpora of English with text from multiple genres, we
quantify the impact of genre diversity in training data for achieving
generalization to text types unseen during training. Our results show that a
heterogeneous training regime is critical for stable and generalizable models,
across parser architectures. We also provide error analyses of model outputs
and out-of-domain performance. To our knowledge, this study is the first to
fully evaluate cross-corpus RST parsing generalizability on complete trees,
examine between-genre degradation within an RST corpus, and investigate the
impact of genre diversity in training data composition. | cs.CL | Why Can't Discourse Parsing Generalize? A Thorough Investigation of the Impact of Data Diversity |
2023-02-13T15:41:20Z | http://arxiv.org/pdf/2302.06461v1 | http://arxiv.org/abs/2302.06461v1 | Kai Shen, Junliang Guo, Xu Tan, Siliang Tang, Rui Wang, Jiang Bian | 2023-02-13T15:41:20Z | cs.CL, cs.AI, cs.LG | The Transformer architecture consists of self-attention and feed-forward
networks (FFNs) which can be viewed as key-value memories according to previous
works. However, FFN and traditional memory utilize different activation
functions (i.e., ReLU and Softmax respectively), which makes them not
equivalent. In this paper, we first rebuild the connections between FFN and
key-value memory by conducting extensive studies on ReLU and Softmax, and find
they are equivalent when adding an additional layer normalization module on
Softmax. In addition, ReLU outperforms Softmax on both FFN and key-value memory
when the number of value slots is large. We analyze the reasons and then
explore this good property of ReLU on the self-attention network where the
original Softmax activation performs poorly on long input sequences. We then
propose a full ReLU architecture named ReLUFormer which performs better than
the baseline Transformer on long sequence tasks such as document translation.
This paper sheds light on the following points: 1) Softmax and ReLU use
different normalization methods over elements which lead to different variances
of results, and ReLU is good at dealing with a large number of key-value slots;
2) FFN and key-value memory are equivalent, and thus the Transformer can be
viewed as a memory network where FFNs and self-attention networks are both
key-value memories. | cs.CL | A Study on ReLU and Softmax in Transformer |
2023-02-13T15:39:08Z | http://arxiv.org/pdf/2302.06459v2 | http://arxiv.org/abs/2302.06459v2 | Lorenzo Lupo, Marco Dinarelli, Laurent Besacier | 2023-04-04T19:05:47Z | cs.CL | Context-aware translation can be achieved by processing a concatenation of
consecutive sentences with the standard Transformer architecture. This paper
investigates the intuitive idea of providing the model with explicit
information about the position of the sentences contained in the concatenation
window. We compare various methods to encode sentence positions into token
representations, including novel methods. Our results show that the Transformer
benefits from certain sentence position encoding methods on English to Russian
translation if trained with a context-discounted loss (Lupo et al., 2022).
However, the same benefits are not observed in English to German. Further
empirical efforts are necessary to define the conditions under which the
proposed approach is beneficial. | cs.CL | Encoding Sentence Position in Context-Aware Neural Machine Translation with Concatenation |
2023-02-13T15:28:02Z | http://arxiv.org/pdf/2302.06448v1 | http://arxiv.org/abs/2302.06448v1 | T. Y. S. S. Santosh, Philipp Bock, Matthias Grabmair | 2023-02-13T15:28:02Z | cs.CL, cs.IR | Segmentation and Rhetorical Role Labeling of legal judgements play a crucial
role in retrieval and adjacent tasks, including case summarization, semantic
search, argument mining etc. Previous approaches have formulated this task
either as independent classification or sequence labeling of sentences. In this
work, we reformulate the task at span level as identifying spans of multiple
consecutive sentences that share the same rhetorical role label to be assigned
via classification. We employ semi-Markov Conditional Random Fields (CRF) to
jointly learn span segmentation and span label assignment. We further explore
three data augmentation strategies to mitigate the data scarcity in the
specialized domain of law where individual documents tend to be very long and
annotation cost is high. Our experiments demonstrate improvement of span-level
prediction metrics with a semi-Markov CRF model over a CRF baseline. This
benefit is contingent on the presence of multi sentence spans in the document. | cs.CL | Joint Span Segmentation and Rhetorical Role Labeling with Data Augmentation for Legal Documents |
2023-02-13T15:18:27Z | http://arxiv.org/pdf/2302.10282v1 | http://arxiv.org/abs/2302.10282v1 | Henrik Voigt, Jan Hombeck, Monique Meuschke, Kai Lawonn, Sina Zarrieß | 2023-02-13T15:18:27Z | cs.CV, cs.AI, cs.CL, cs.LG | Existing language and vision models achieve impressive performance in
image-text understanding. Yet, it is an open question to what extent they can
be used for language understanding in 3D environments and whether they
implicitly acquire 3D object knowledge, e.g. about different views of an
object. In this paper, we investigate whether a state-of-the-art language and
vision model, CLIP, is able to ground perspective descriptions of a 3D object
and identify canonical views of common objects based on text queries. We
present an evaluation framework that uses a circling camera around a 3D object
to generate images from different viewpoints and evaluate them in terms of
their similarity to natural language descriptions. We find that a pre-trained
CLIP model performs poorly on most canonical views and that fine-tuning using
hard negative sampling and random contrasting yields good results even under
conditions with little available training data. | cs.CV | Paparazzi: A Deep Dive into the Capabilities of Language and Vision Models for Grounding Viewpoint Descriptions |
2023-02-13T15:03:07Z | http://arxiv.org/pdf/2302.06426v2 | http://arxiv.org/abs/2302.06426v2 | Miguel Ortega-Martín, Óscar García-Sierra, Alfonso Ardoiz, Jorge Álvarez, Juan Carlos Armenteros, Adrián Alonso | 2023-02-20T08:55:39Z | cs.CL, cs.AI | Linguistic ambiguity is and has always been one of the main challenges in
Natural Language Processing (NLP) systems. Modern Transformer architectures
like BERT, T5 or more recently InstructGPT have achieved some impressive
improvements in many NLP fields, but there is still plenty of work to do.
Motivated by the uproar caused by ChatGPT, in this paper we provide an
introduction to linguistic ambiguity, its varieties and their relevance in
modern NLP, and perform an extensive empiric analysis. ChatGPT strengths and
weaknesses are revealed, as well as strategies to get the most of this model. | cs.CL | Linguistic ambiguity analysis in ChatGPT |
2023-02-13T14:36:21Z | http://arxiv.org/pdf/2302.06397v2 | http://arxiv.org/abs/2302.06397v2 | Yongqi Li, Yu Yu, Tieyun Qian | 2023-10-16T12:31:12Z | cs.CL | Despite the recent success achieved by several two-stage prototypical
networks in few-shot named entity recognition (NER) task, the overdetected
false spans at the span detection stage and the inaccurate and unstable
prototypes at the type classification stage remain to be challenging problems.
In this paper, we propose a novel Type-Aware Decomposed framework, namely
TadNER, to solve these problems. We first present a type-aware span filtering
strategy to filter out false spans by removing those semantically far away from
type names. We then present a type-aware contrastive learning strategy to
construct more accurate and stable prototypes by jointly exploiting support
samples and type names as references. Extensive experiments on various
benchmarks prove that our proposed TadNER framework yields a new
state-of-the-art performance. Our code and data will be available at
https://github.com/NLPWM-WHU/TadNER. | cs.CL | Type-Aware Decomposed Framework for Few-Shot Named Entity Recognition |
2023-02-13T13:24:11Z | http://arxiv.org/pdf/2302.06341v1 | http://arxiv.org/abs/2302.06341v1 | Tianshuo Zang, Maolin Yang, Wentao Yong, Pingyu Jiang | 2023-02-13T13:24:11Z | cs.AI, cs.CL, cs.IR | Retrieving the similar solutions from the historical case base for new design
requirements is the first step in mechanical part redesign under the context of
case-based reasoning. However, the manual retrieving method has the problem of
low efficiency when the case base is large. Additionally, it is difficult for
simple reasoning algorithms (e.g., rule-based reasoning, decision tree) to
cover all the features in complicated design solutions. In this regard, a
text2shape deep retrieval model is established in order to support text
description-based mechanical part shapes retrieval, where the texts are for
describing the structural features of the target mechanical parts. More
specifically, feature engineering is applied to identify the key structural
features of the target mechanical parts. Based on the identified key structural
features, a training set of 1000 samples was constructed, where each sample
consisted of a paragraph of text description of a group of structural features
and the corresponding 3D shape of the structural features. RNN and 3D CNN
algorithms were customized to build the text2shape deep retrieval model.
Orthogonal experiments were used for modeling turning. Eventually, the highest
accuracy of the model was 0.98; therefore, the model can be effective for
retrieving initial cases for mechanical part redesign. | cs.AI | Text2shape Deep Retrieval Model: Generating Initial Cases for Mechanical Part Redesign under the Context of Case-Based Reasoning |
2023-02-13T13:14:23Z | http://arxiv.org/pdf/2302.06337v3 | http://arxiv.org/abs/2302.06337v3 | Zhijun Chen, Hailong Sun, Haoqian He, Pengpeng Chen | 2023-03-19T13:01:47Z | cs.LG, cs.AI, cs.CL, cs.HC | This paper explores the integration of symbolic logic knowledge into deep
neural networks for learning from noisy crowd labels. We introduce Logic-guided
Learning from Noisy Crowd Labels (Logic-LNCL), an EM-alike iterative logic
knowledge distillation framework that learns from both noisy labeled data and
logic rules of interest. Unlike traditional EM methods, our framework contains
a ``pseudo-E-step'' that distills from the logic rules a new type of learning
target, which is then used in the ``pseudo-M-step'' for training the
classifier. Extensive evaluations on two real-world datasets for text sentiment
classification and named entity recognition demonstrate that the proposed
framework improves the state-of-the-art and provides a new solution to learning
from noisy crowd labels. | cs.LG | Learning from Noisy Crowd Labels with Logics |
2023-02-13T12:39:45Z | http://arxiv.org/pdf/2302.06321v2 | http://arxiv.org/abs/2302.06321v2 | Deepak Kumar, Oleg Lesota, George Zerveas, Daniel Cohen, Carsten Eickhoff, Markus Schedl, Navid Rekabsaz | 2023-06-18T20:14:15Z | cs.CL, cs.AI | Large pre-trained language models contain societal biases and carry along
these biases to downstream tasks. Current in-processing bias mitigation
approaches (like adversarial training) impose debiasing by updating a model's
parameters, effectively transferring the model to a new, irreversible debiased
state. In this work, we propose a novel approach to develop stand-alone
debiasing functionalities separate from the model, which can be integrated into
the model on-demand, while keeping the core model untouched. Drawing from the
concept of AdapterFusion in multi-task learning, we introduce DAM (Debiasing
with Adapter Modules) - a debiasing approach to first encapsulate arbitrary
bias mitigation functionalities into separate adapters, and then add them to
the model on-demand in order to deliver fairness qualities. We conduct a large
set of experiments on three classification tasks with gender, race, and age as
protected attributes. Our results show that DAM improves or maintains the
effectiveness of bias mitigation, avoids catastrophic forgetting in a
multi-attribute scenario, and maintains on-par task performance, while granting
parameter-efficiency and easy switching between the original and debiased
models. | cs.CL | Parameter-efficient Modularised Bias Mitigation via AdapterFusion |
2023-02-13T09:15:00Z | http://arxiv.org/pdf/2302.06198v3 | http://arxiv.org/abs/2302.06198v3 | Hongjing Li, Hanqi Yan, Yanran Li, Li Qian, Yulan He, Lin Gui | 2023-05-10T09:16:53Z | cs.CL | Recent years have witnessed increasing interests in prompt-based learning in
which models can be trained on only a few annotated instances, making them
suitable in low-resource settings. When using prompt-based learning for text
classification, the goal is to use a pre-trained language model (PLM) to
predict a missing token in a pre-defined template given an input text, which
can be mapped to a class label. However, PLMs built on the transformer
architecture tend to generate similar output embeddings, making it difficult to
discriminate between different class labels. The problem is further exacerbated
when dealing with classification tasks involving many fine-grained class
labels. In this work, we alleviate this information diffusion issue, i.e.,
different tokens share a large proportion of similar information after going
through stacked multiple self-attention layers in a transformer, by proposing a
calibration method built on feature transformations through rotation and
scaling to map a PLM-encoded embedding into a new metric space to guarantee the
distinguishability of the resulting embeddings. Furthermore, we take the
advantage of hyperbolic embeddings to capture the hierarchical relations among
fine-grained class-associated token embedding by a coarse-to-fine metric
learning strategy to enhance the distinguishability of the learned output
embeddings. Extensive experiments on the three datasets under various settings
demonstrate the effectiveness of our approach. Our code can be found at
https://github.com/donttal/TARA. | cs.CL | Distinguishability Calibration to In-Context Learning |
2023-02-13T08:23:03Z | http://arxiv.org/pdf/2302.06174v1 | http://arxiv.org/abs/2302.06174v1 | Ricardo Schiffers, Dagmar Kern, Daniel Hienert | 2023-02-13T08:23:03Z | cs.CL | Word embeddings are an essential instrument in many NLP tasks. Most available
resources are trained on general language from Web corpora or Wikipedia dumps.
However, word embeddings for domain-specific language are rare, in particular
for the social science domain. Therefore, in this work, we describe the
creation and evaluation of word embedding models based on 37,604 open-access
social science research papers. In the evaluation, we compare domain-specific
and general language models for (i) language coverage, (ii) diversity, and
(iii) semantic relationships. We found that the created domain-specific model,
even with a relatively small vocabulary size, covers a large part of social
science concepts, their neighborhoods are diverse in comparison to more general
models. Across all relation types, we found a more extensive coverage of
semantic relationships. | cs.CL | Evaluation of Word Embeddings for the Social Sciences |
2023-02-13T07:33:46Z | http://arxiv.org/pdf/2302.06155v1 | http://arxiv.org/abs/2302.06155v1 | Shashank Mujumdar, Stuti Mehta, Hima Patel, Suman Mitra | 2023-02-13T07:33:46Z | cs.CL, cs.AI | In this paper, we investigate the effect of addressing difficult samples from
a given text dataset on the downstream text classification task. We define
difficult samples as being non-obvious cases for text classification by
analysing them in the semantic embedding space; specifically - (i) semantically
similar samples that belong to different classes and (ii) semantically
dissimilar samples that belong to the same class. We propose a penalty function
to measure the overall difficulty score of every sample in the dataset. We
conduct exhaustive experiments on 13 standard datasets to show a consistent
improvement of up to 9% and discuss qualitative results to show effectiveness
of our approach in identifying difficult samples for a text classification
model. | cs.CL | Identifying Semantically Difficult Samples to Improve Text Classification |
2023-02-13T06:38:25Z | http://arxiv.org/pdf/2302.06132v3 | http://arxiv.org/abs/2302.06132v3 | Irene Li, Boming Yang | 2023-10-19T07:47:44Z | cs.CL | Knowledge graph completion (KGC) aims to discover missing relations of query
entities. Current text-based models utilize the entity name and description to
infer the tail entity given the head entity and a certain relation. Existing
approaches also consider the neighborhood of the head entity. However, these
methods tend to model the neighborhood using a flat structure and are only
restricted to 1-hop neighbors. In this work, we propose a node
neighborhood-enhanced framework for knowledge graph completion. It models the
head entity neighborhood from multiple hops using graph neural networks to
enrich the head node information. Moreover, we introduce an additional edge
link prediction task to improve KGC. Evaluation on two public datasets shows
that this framework is simple yet effective. The case study also shows that the
model is able to predict explainable predictions. | cs.CL | NNKGC: Improving Knowledge Graph Completion with Node Neighborhoods |
2023-02-13T04:56:11Z | http://arxiv.org/pdf/2302.06100v2 | http://arxiv.org/abs/2302.06100v2 | Andrew Blair-Stanek, Nils Holzenberger, Benjamin Van Durme | 2023-05-10T19:17:23Z | cs.CL, cs.AI | Statutory reasoning is the task of reasoning with facts and statutes, which
are rules written in natural language by a legislature. It is a basic legal
skill. In this paper we explore the capabilities of the most capable GPT-3
model, text-davinci-003, on an established statutory-reasoning dataset called
SARA. We consider a variety of approaches, including dynamic few-shot
prompting, chain-of-thought prompting, and zero-shot prompting. While we
achieve results with GPT-3 that are better than the previous best published
results, we also identify several types of clear errors it makes. We
investigate why these errors happen. We discover that GPT-3 has imperfect prior
knowledge of the actual U.S. statutes on which SARA is based. More importantly,
we create simple synthetic statutes, which GPT-3 is guaranteed not to have seen
during training. We find GPT-3 performs poorly at answering straightforward
questions about these simple synthetic statutes. | cs.CL | Can GPT-3 Perform Statutory Reasoning? |
2023-02-13T03:25:37Z | http://arxiv.org/pdf/2302.06078v2 | http://arxiv.org/abs/2302.06078v2 | Yu-Chien Tang, Kuang-Da Wang, Ting-Yun Ou, Wen-Chih Peng | 2023-02-14T03:53:02Z | cs.CL, cs.CV | This paper presents a robust solution to the Memotion 3.0 Shared Task. The
goal of this task is to classify the emotion and the corresponding intensity
expressed by memes, which are usually in the form of images with short captions
on social media. Understanding the multi-modal features of the given memes will
be the key to solving the task. In this work, we use CLIP to extract aligned
image-text features and propose a novel meme sentiment analysis framework,
consisting of a Cooperative Teaching Model (CTM) for Task A and a Cascaded
Emotion Classifier (CEC) for Tasks B&C. CTM is based on the idea of knowledge
distillation, and can better predict the sentiment of a given meme in Task A;
CEC can leverage the emotion intensity suggestion from the prediction of Task C
to classify the emotion more precisely in Task B. Experiments show that we
achieved the 2nd place ranking for both Task A and Task B and the 4th place
ranking for Task C, with weighted F1-scores of 0.342, 0.784, and 0.535
respectively. The results show the robustness and effectiveness of our
framework. Our code is released at github. | cs.CL | NYCU-TWO at Memotion 3: Good Foundation, Good Teacher, then you have Good Meme Analysis |
2023-02-13T01:49:51Z | http://arxiv.org/pdf/2302.07727v2 | http://arxiv.org/abs/2302.07727v2 | Bingyan Wang, Rongpeng Li, Jianhang Zhu, Zhifeng Zhao, Honggang Zhang | 2023-04-15T10:25:24Z | cs.CL, cs.LG | In recent years, with the rapid development of deep learning and natural
language processing technologies, semantic communication has become a topic of
great interest in the field of communication. Although existing deep
learning-based semantic communication approaches have shown many advantages,
they still do not make sufficient use of prior knowledge. Moreover, most
existing semantic communication methods focus on the semantic encoding at the
transmitter side, while we believe that the semantic decoding capability of the
receiver should also be concerned. In this paper, we propose a knowledge
enhanced semantic communication framework in which the receiver can more
actively utilize the facts in the knowledge base for semantic reasoning and
decoding, on the basis of only affecting the parameters rather than the
structure of the neural networks at the transmitter side. Specifically, we
design a transformer-based knowledge extractor to find relevant factual triples
for the received noisy signal. Extensive simulation results on the WebNLG
dataset demonstrate that the proposed receiver yields superior performance on
top of the knowledge graph enhanced decoding. | cs.CL | Knowledge Enhanced Semantic Communication Receiver |
2023-02-13T00:06:47Z | http://arxiv.org/pdf/2302.06029v1 | http://arxiv.org/abs/2302.06029v1 | Xiaochen Zhang, Daniel Tang | 2023-02-13T00:06:47Z | cs.CL, cs.LG | We leverage different context windows when predicting the emotion of
different utterances. New modules are included to realize variable-length
context: 1) two speaker-aware units, which explicitly model inner- and
inter-speaker dependencies to form distilled conversational context, and 2) a
top-k normalization layer, which determines the most proper context windows
from the conversational context to predict emotion. Experiments and ablation
studies show that our approach outperforms several strong baselines on three
public datasets. | cs.CL | Emotion Detection in Unfix-length-Context Conversation |
2023-02-12T22:05:08Z | http://arxiv.org/pdf/2304.11079v1 | http://arxiv.org/abs/2304.11079v1 | Oğuz 'Oz' Buruk | 2023-02-12T22:05:08Z | cs.CL, cs.AI, cs.HC, cs.LG | The debate around the use of GPT 3.5 has been a popular topic among academics
since the release of ChatGPT. Whilst some have argued for the advantages of GPT
3.5 in enhancing academic writing, others have raised concerns such as
plagiarism, the spread of false information, and ecological issues. The need
for finding ways to use GPT 3.5 models transparently has been voiced, and
suggestions have been made on social media as to how to use GPT 3.5 models in a
smart way. Nevertheless, to date, there is a lack of literature which clearly
outlines how to use GPT 3.5 models in academic writing, how effective they are,
and how to use them transparently. To address this, I conducted a personal
experience experiment with GPT 3.5, specifically by using OpenAI text davinci
003 model, for writing this article. I identified five ways of using GPT 3.5:
Chunk Stylist, Bullet to Paragraph, Talk Textualizer, Research Buddy, and
Polisher. I reflected on their efficacy, and commented on their potential
impact on writing ethics. Additionally, I provided a comprehensive document
which shows the prompts I used, results I got from GPT 3.5, the final edits and
visually compares those by showing the differences in percentages. | cs.CL | Academic Writing with GPT-3.5: Reflections on Practices, Efficacy and Transparency |
2023-02-12T21:45:18Z | http://arxiv.org/pdf/2302.06008v1 | http://arxiv.org/abs/2302.06008v1 | Johannes Wirth, René Peinl | 2023-02-12T21:45:18Z | cs.CL, cs.AI, cs.LG, cs.SD, eess.AS | We present ASR Bundestag, a dataset for automatic speech recognition in
German, consisting of 610 hours of aligned audio-transcript pairs for
supervised training as well as 1,038 hours of unlabeled audio snippets for
self-supervised learning, based on raw audio data and transcriptions from
plenary sessions and committee meetings of the German parliament. In addition,
we discuss utilized approaches for the automated creation of speech datasets
and assess the quality of the resulting dataset based on evaluations and
finetuning of a pre-trained state of the art model. We make the dataset
publicly available, including all subsets. | cs.CL | ASR Bundestag: A Large-Scale political debate dataset in German |
2023-02-12T20:36:55Z | http://arxiv.org/pdf/2302.10281v1 | http://arxiv.org/abs/2302.10281v1 | Andre Nakkab, Benjamin Feuer, Chinmay Hegde | 2023-02-12T20:36:55Z | cs.CV, cs.AI, cs.CL | Recent advances in training vision-language models have demonstrated
unprecedented robustness and transfer learning effectiveness; however, standard
computer vision datasets are image-only, and therefore not well adapted to such
training methods. Our paper introduces a simple methodology for adapting any
fine-grained image classification dataset for distributed vision-language
pretraining. We implement this methodology on the challenging iNaturalist-2021
dataset, comprised of approximately 2.7 million images of macro-organisms
across 10,000 classes, and achieve a new state-of-the art model in terms of
zero-shot classification accuracy. Somewhat surprisingly, our model (trained
using a new method called locked-image text tuning) uses a pre-trained, frozen
vision representation, proving that language alignment alone can attain strong
transfer learning performance, even on fractious, long-tailed datasets. Our
approach opens the door for utilizing high quality vision-language pretrained
models in agriculturally relevant applications involving species detection. | cs.CV | LiT Tuned Models for Efficient Species Detection |
2023-02-12T19:12:24Z | http://arxiv.org/pdf/2302.05981v3 | http://arxiv.org/abs/2302.05981v3 | Shyam Sudhakaran, Miguel González-Duque, Claire Glanois, Matthias Freiberger, Elias Najarro, Sebastian Risi | 2023-11-08T09:43:51Z | cs.AI, cs.CL, cs.LG | Procedural Content Generation (PCG) is a technique to generate complex and
diverse environments in an automated way. However, while generating content
with PCG methods is often straightforward, generating meaningful content that
reflects specific intentions and constraints remains challenging. Furthermore,
many PCG algorithms lack the ability to generate content in an open-ended
manner. Recently, Large Language Models (LLMs) have shown to be incredibly
effective in many diverse domains. These trained LLMs can be fine-tuned,
re-using information and accelerating training for new tasks. Here, we
introduce MarioGPT, a fine-tuned GPT2 model trained to generate tile-based game
levels, in our case Super Mario Bros levels. MarioGPT can not only generate
diverse levels, but can be text-prompted for controllable level generation,
addressing one of the key challenges of current PCG techniques. As far as we
know, MarioGPT is the first text-to-level model and combined with novelty
search it enables the generation of diverse levels with varying play-style
dynamics (i.e. player paths) and the open-ended discovery of an increasingly
diverse range of content. Code available at
https://github.com/shyamsn97/mario-gpt. | cs.AI | MarioGPT: Open-Ended Text2Level Generation through Large Language Models |
2023-02-12T18:08:54Z | http://arxiv.org/pdf/2302.07740v1 | http://arxiv.org/abs/2302.07740v1 | Wei-Wei Du, Hong-Wei Wu, Wei-Yao Wang, Wen-Chih Peng | 2023-02-12T18:08:54Z | cs.CL, cs.AI, cs.CV, cs.LG | Multi-modal fact verification has become an important but challenging issue
on social media due to the mismatch between the text and images in the
misinformation of news content, which has been addressed by considering
cross-modalities to identify the veracity of the news in recent years. In this
paper, we propose the Pre-CoFactv2 framework with new parameter-efficient
foundation models for modeling fine-grained text and input embeddings with
lightening parameters, multi-modal multi-type fusion for not only capturing
relations for the same and different modalities but also for different types
(i.e., claim and document), and feature representations for explicitly
providing metadata for each sample. In addition, we introduce a unified
ensemble method to boost model performance by adjusting the importance of each
trained model with not only the weights but also the powers. Extensive
experiments show that Pre-CoFactv2 outperforms Pre-CoFact by a large margin and
achieved new state-of-the-art results at the Factify challenge at AAAI 2023. We
further illustrate model variations to verify the relative contributions of
different components. Our team won the first prize (F1-score: 81.82%) and we
made our code publicly available at
https://github.com/wwweiwei/Pre-CoFactv2-AAAI-2023. | cs.CL | Team Triple-Check at Factify 2: Parameter-Efficient Large Foundation Models with Feature Representations for Multi-Modal Fact Verification |
2023-02-12T17:41:16Z | http://arxiv.org/pdf/2302.05965v3 | http://arxiv.org/abs/2302.05965v3 | Haoyang Li, Jing Zhang, Cuiping Li, Hong Chen | 2023-04-10T14:58:49Z | cs.CL | One of the recent best attempts at Text-to-SQL is the pre-trained language
model. Due to the structural property of the SQL queries, the seq2seq model
takes the responsibility of parsing both the schema items (i.e., tables and
columns) and the skeleton (i.e., SQL keywords). Such coupled targets increase
the difficulty of parsing the correct SQL queries especially when they involve
many schema items and logic operators. This paper proposes a ranking-enhanced
encoding and skeleton-aware decoding framework to decouple the schema linking
and the skeleton parsing. Specifically, for a seq2seq encoder-decode model, its
encoder is injected by the most relevant schema items instead of the whole
unordered ones, which could alleviate the schema linking effort during SQL
parsing, and its decoder first generates the skeleton and then the actual SQL
query, which could implicitly constrain the SQL parsing. We evaluate our
proposed framework on Spider and its three robustness variants: Spider-DK,
Spider-Syn, and Spider-Realistic. The experimental results show that our
framework delivers promising performance and robustness. Our code is available
at https://github.com/RUCKBReasoning/RESDSQL. | cs.CL | RESDSQL: Decoupling Schema Linking and Skeleton Parsing for Text-to-SQL |
2023-02-12T17:32:55Z | http://arxiv.org/pdf/2302.05963v1 | http://arxiv.org/abs/2302.05963v1 | Xanh Ho, Anh-Khoa Duong Nguyen, Saku Sugawara, Akiko Aizawa | 2023-02-12T17:32:55Z | cs.CL | To explain the predicted answers and evaluate the reasoning abilities of
models, several studies have utilized underlying reasoning (UR) tasks in
multi-hop question answering (QA) datasets. However, it remains an open
question as to how effective UR tasks are for the QA task when training models
on both tasks in an end-to-end manner. In this study, we address this question
by analyzing the effectiveness of UR tasks (including both sentence-level and
entity-level tasks) in three aspects: (1) QA performance, (2) reasoning
shortcuts, and (3) robustness. While the previous models have not been
explicitly trained on an entity-level reasoning prediction task, we build a
multi-task model that performs three tasks together: sentence-level supporting
facts prediction, entity-level reasoning prediction, and answer prediction.
Experimental results on 2WikiMultiHopQA and HotpotQA-small datasets reveal that
(1) UR tasks can improve QA performance. Using four debiased datasets that are
newly created, we demonstrate that (2) UR tasks are helpful in preventing
reasoning shortcuts in the multi-hop QA task. However, we find that (3) UR
tasks do not contribute to improving the robustness of the model on adversarial
questions, such as sub-questions and inverted questions. We encourage future
studies to investigate the effectiveness of entity-level reasoning in the form
of natural language questions (e.g., sub-question forms). | cs.CL | Analyzing the Effectiveness of the Underlying Reasoning Tasks in Multi-hop Question Answering |
2023-02-12T15:05:10Z | http://arxiv.org/pdf/2302.05932v1 | http://arxiv.org/abs/2302.05932v1 | Derek Chen, Kun Qian, Zhou Yu | 2023-02-12T15:05:10Z | cs.CL | Prompt-based methods with large pre-trained language models (PLMs) have shown
impressive unaided performance across many NLP tasks. These models improve even
further with the addition of a few labeled in-context exemplars to guide output
generation. However, for more complex tasks such as dialogue state tracking
(DST), designing prompts that reliably convey the desired intent is nontrivial,
leading to unstable results. Furthermore, building in-context exemplars for
dialogue tasks is difficult because conversational contexts are long while
model input lengths are relatively short. To overcome these issues we first
adapt a meta-learning scheme to the dialogue domain which stabilizes the
ability of the model to perform well under various prompts. We additionally
design a novel training method to improve upon vanilla retrieval mechanisms to
find ideal in-context examples. Finally, we introduce a saliency model to limit
dialogue text length, allowing us to include more exemplars per query. In
effect, we are able to achieve highly competitive results for few-shot DST on
MultiWOZ. | cs.CL | Stabilized In-Context Learning with Pre-trained Language Models for Few Shot Dialogue State Tracking |
2023-02-12T13:31:53Z | http://arxiv.org/pdf/2302.05913v1 | http://arxiv.org/abs/2302.05913v1 | Stuart Mesham, Christopher Bryant, Marek Rei, Zheng Yuan | 2023-02-12T13:31:53Z | cs.CL | We extend a current sequence-tagging approach to Grammatical Error Correction
(GEC) by introducing specialised tags for spelling correction and morphological
inflection using the SymSpell and LemmInflect algorithms. Our approach improves
generalisation: the proposed new tagset allows a smaller number of tags to
correct a larger range of errors. Our results show a performance improvement
both overall and in the targeted error categories. We further show that
ensembles trained with our new tagset outperform those trained with the
baseline tagset on the public BEA benchmark. | cs.CL | An Extended Sequence Tagging Vocabulary for Grammatical Error Correction |
2023-02-12T12:43:36Z | http://arxiv.org/pdf/2302.05900v1 | http://arxiv.org/abs/2302.05900v1 | Sebastien Montella, Alexis Nasr, Johannes Heinecke, Frederic Bechet, Lina M. Rojas-Barahona | 2023-02-12T12:43:36Z | cs.CL | Text generation from Abstract Meaning Representation (AMR) has substantially
benefited from the popularized Pretrained Language Models (PLMs). Myriad
approaches have linearized the input graph as a sequence of tokens to fit the
PLM tokenization requirements. Nevertheless, this transformation jeopardizes
the structural integrity of the graph and is therefore detrimental to its
resulting representation. To overcome this issue, Ribeiro et al. have recently
proposed StructAdapt, a structure-aware adapter which injects the input graph
connectivity within PLMs using Graph Neural Networks (GNNs). In this paper, we
investigate the influence of Relative Position Embeddings (RPE) on AMR-to-Text,
and, in parallel, we examine the robustness of StructAdapt. Through ablation
studies, graph attack and link prediction, we reveal that RPE might be
partially encoding input graphs. We suggest further research regarding the role
of RPE will provide valuable insights for Graph-to-Text generation. | cs.CL | Investigating the Effect of Relative Positional Embeddings on AMR-to-Text Generation with Structural Adapters |
2023-02-12T12:18:01Z | http://arxiv.org/pdf/2302.05898v1 | http://arxiv.org/abs/2302.05898v1 | Diederik Aerts, Jonito Aerts Argëlles, Lester Beltran, Suzette Geriente, Sandro Sozzo | 2023-02-12T12:18:01Z | q-bio.NC, cs.CL, quant-ph | In physics, entanglement 'reduces' the entropy of an entity, because the (von
Neumann) entropy of, e.g., a composite bipartite entity in a pure entangled
state is systematically lower than the entropy of the component sub-entities.
We show here that this 'genuinely non-classical reduction of entropy as a
result of composition' also holds whenever two concepts combine in human
cognition and, more generally, it is valid in human culture. We exploit these
results and make a 'new hypothesis' on the nature of entanglement, namely, the
production of entanglement in the preparation of a composite entity can be seen
as a 'dynamical process of collaboration between its sub-entities to reduce
uncertainty', because the composite entity is in a pure state while its
sub-entities are in a non-pure, or density, state, as a result of the
preparation. We identify within the nature of this entanglement a mechanism of
contextual updating and illustrate the mechanism in the example we analyze. Our
hypothesis naturally explains the 'non-classical nature' of some quantum
logical connectives, as due to Bell-type correlations. | q-bio.NC | Entanglement as a Method to Reduce Uncertainty |
2023-02-12T11:26:10Z | http://arxiv.org/pdf/2302.05895v2 | http://arxiv.org/abs/2302.05895v2 | Chuyuan Li, Patrick Huber, Wen Xiao, Maxime Amblard, Chloé Braud, Giuseppe Carenini | 2023-06-25T10:19:00Z | cs.CL | Discourse processing suffers from data sparsity, especially for dialogues. As
a result, we explore approaches to build discourse structures for dialogues,
based on attention matrices from Pre-trained Language Models (PLMs). We
investigate multiple tasks for fine-tuning and show that the dialogue-tailored
Sentence Ordering task performs best. To locate and exploit discourse
information in PLMs, we propose an unsupervised and a semi-supervised method.
Our proposals achieve encouraging results on the STAC corpus, with F1 scores of
57.2 and 59.3 for unsupervised and semi-supervised methods, respectively. When
restricted to projective trees, our scores improved to 63.3 and 68.1. | cs.CL | Discourse Structure Extraction from Pre-Trained and Fine-Tuned Language Models in Dialogues |
2023-02-12T11:12:44Z | http://arxiv.org/pdf/2302.05892v1 | http://arxiv.org/abs/2302.05892v1 | Lujia Shen, Xuhong Zhang, Shouling Ji, Yuwen Pu, Chunpeng Ge, Xing Yang, Yanghe Feng | 2023-02-12T11:12:44Z | cs.CL, cs.AI, cs.CR | Currently, natural language processing (NLP) models are wildly used in
various scenarios. However, NLP models, like all deep models, are vulnerable to
adversarially generated text. Numerous works have been working on mitigating
the vulnerability from adversarial attacks. Nevertheless, there is no
comprehensive defense in existing works where each work targets a specific
attack category or suffers from the limitation of computation overhead,
irresistible to adaptive attack, etc.
In this paper, we exhaustively investigate the adversarial attack algorithms
in NLP, and our empirical studies have discovered that the attack algorithms
mainly disrupt the importance distribution of words in a text. A well-trained
model can distinguish subtle importance distribution differences between clean
and adversarial texts. Based on this intuition, we propose TextDefense, a new
adversarial example detection framework that utilizes the target model's
capability to defend against adversarial attacks while requiring no prior
knowledge. TextDefense differs from previous approaches, where it utilizes the
target model for detection and thus is attack type agnostic. Our extensive
experiments show that TextDefense can be applied to different architectures,
datasets, and attack methods and outperforms existing methods. We also discover
that the leading factor influencing the performance of TextDefense is the
target model's generalizability. By analyzing the property of the target model
and the property of the adversarial example, we provide our insights into the
adversarial attacks in NLP and the principles of our defense method. | cs.CL | TextDefense: Adversarial Text Detection based on Word Importance Entropy |
2023-02-12T10:13:00Z | http://arxiv.org/pdf/2302.05888v1 | http://arxiv.org/abs/2302.05888v1 | Hsuan Su, Shachi H Kumar, Sahisnu Mazumder, Wenda Chen, Ramesh Manuvinakurike, Eda Okur, Saurav Sahay, Lama Nachman, Shang-Tse Chen, Hung-yi Lee | 2023-02-12T10:13:00Z | cs.CL, cs.AI, cs.LG | With the power of large pretrained language models, various research works
have integrated knowledge into dialogue systems. The traditional techniques
treat knowledge as part of the input sequence for the dialogue system,
prepending a set of knowledge statements in front of dialogue history. However,
such a mechanism forces knowledge sets to be concatenated in an ordered manner,
making models implicitly pay imbalanced attention to the sets during training.
In this paper, we first investigate how the order of the knowledge set can
influence autoregressive dialogue systems' responses. We conduct experiments on
two commonly used dialogue datasets with two types of transformer-based models
and find that models view the input knowledge unequally. To this end, we
propose a simple and novel technique to alleviate the order effect by modifying
the position embeddings of knowledge input in these models. With the proposed
position embedding method, the experimental results show that each knowledge
statement is uniformly considered to generate responses. | cs.CL | Position Matters! Empirical Study of Order Effect in Knowledge-grounded Dialogue |
2023-02-12T07:04:55Z | http://arxiv.org/pdf/2302.07728v1 | http://arxiv.org/abs/2302.07728v1 | Guangyi Xiao, Xinlong Liu, Hao Chen, Jingzhi Guo, Zhiguo Gong | 2023-02-12T07:04:55Z | cs.CL, cs.AI | In this paper, we study the problem of legal domain adaptation problem from
an imbalanced source domain to a partial target domain. The task aims to
improve legal judgment predictions for non-professional fact descriptions. We
formulate this task as a partial-and-imbalanced domain adaptation problem.
Though deep domain adaptation has achieved cutting-edge performance in many
unsupervised domain adaptation tasks. However, due to the negative transfer of
samples in non-shared classes, it is hard for current domain adaptation model
to solve the partial-and-imbalanced transfer problem. In this work, we explore
large-scale non-shared but related classes data in the source domain with a
hierarchy weighting adaptation to tackle this limitation. We propose to embed a
novel pArtial Imbalanced Domain Adaptation technique (AIDA) in the deep
learning model, which can jointly borrow sibling knowledge from non-shared
classes to shared classes in the source domain and further transfer the shared
classes knowledge from the source domain to the target domain. Experimental
results show that our model outperforms the state-of-the-art algorithms. | cs.CL | AIDA: Legal Judgment Predictions for Non-Professional Fact Descriptions via Partial-and-Imbalanced Domain Adaptation |
2023-02-12T04:21:49Z | http://arxiv.org/pdf/2302.05852v1 | http://arxiv.org/abs/2302.05852v1 | Jiaming Shen, Jialu Liu, Dan Finnie, Negar Rahmati, Michael Bendersky, Marc Najork | 2023-02-12T04:21:49Z | cs.CL, cs.AI, cs.IR | Automatic headline generation enables users to comprehend ongoing news events
promptly and has recently become an important task in web mining and natural
language processing. With the growing need for news headline generation, we
argue that the hallucination issue, namely the generated headlines being not
supported by the original news stories, is a critical challenge for the
deployment of this feature in web-scale systems Meanwhile, due to the
infrequency of hallucination cases and the requirement of careful reading for
raters to reach the correct consensus, it is difficult to acquire a large
dataset for training a model to detect such hallucinations through human
curation. In this work, we present a new framework named ExHalder to address
this challenge for headline hallucination detection. ExHalder adapts the
knowledge from public natural language inference datasets into the news domain
and learns to generate natural language sentences to explain the hallucination
detection results. To evaluate the model performance, we carefully collect a
dataset with more than six thousand labeled <article, headline> pairs.
Extensive experiments on this dataset and another six public ones demonstrate
that ExHalder can identify hallucinated headlines accurately and justifies its
predictions with human-readable natural language explanations. | cs.CL | "Why is this misleading?": Detecting News Headline Hallucinations with Explanations |
2023-02-12T01:26:49Z | http://arxiv.org/pdf/2302.07730v4 | http://arxiv.org/abs/2302.07730v4 | Xavier Amatriain, Ananth Sankar, Jie Bing, Praveen Kumar Bodigutla, Timothy J. Hazen, Michaeel Kazi | 2024-03-31T21:20:30Z | cs.CL | In the past few years we have seen the meteoric appearance of dozens of
foundation models of the Transformer family, all of which have memorable and
sometimes funny, but not self-explanatory, names. The goal of this paper is to
offer a somewhat comprehensive but simple catalog and classification of the
most popular Transformer models. The paper also includes an introduction to the
most important aspects and innovations in Transformer models. Our catalog will
include models that are trained using self-supervised learning (e.g., BERT or
GPT3) as well as those that are further trained using a human-in-the-loop (e.g.
the InstructGPT model used by ChatGPT). | cs.CL | Transformer models: an introduction and catalog |
2023-02-11T23:34:42Z | http://arxiv.org/pdf/2302.05817v2 | http://arxiv.org/abs/2302.05817v2 | Graham Todd, Sam Earle, Muhammad Umair Nasir, Michael Cerny Green, Julian Togelius | 2023-06-01T05:43:02Z | cs.AI, cs.CL, cs.NE | Large Language Models (LLMs) are powerful tools, capable of leveraging their
training on natural language to write stories, generate code, and answer
questions. But can they generate functional video game levels? Game levels,
with their complex functional constraints and spatial relationships in more
than one dimension, are very different from the kinds of data an LLM typically
sees during training. Datasets of game levels are also hard to come by,
potentially taxing the abilities of these data-hungry models. We investigate
the use of LLMs to generate levels for the game Sokoban, finding that LLMs are
indeed capable of doing so, and that their performance scales dramatically with
dataset size. We also perform preliminary experiments on controlling LLM level
generators and discuss promising areas for future work. | cs.AI | Level Generation Through Large Language Models |
2023-02-11T18:51:23Z | http://arxiv.org/pdf/2302.05759v1 | http://arxiv.org/abs/2302.05759v1 | Lee Kezar, Jesse Thomason, Zed Sevcikova Sehyr | 2023-02-11T18:51:23Z | cs.CL, cs.CV, I.2.7; I.2.10; I.5.4 | We use insights from research on American Sign Language (ASL) phonology to
train models for isolated sign language recognition (ISLR), a step towards
automatic sign language understanding. Our key insight is to explicitly
recognize the role of phonology in sign production to achieve more accurate
ISLR than existing work which does not consider sign language phonology. We
train ISLR models that take in pose estimations of a signer producing a single
sign to predict not only the sign but additionally its phonological
characteristics, such as the handshape. These auxiliary predictions lead to a
nearly 9% absolute gain in sign recognition accuracy on the WLASL benchmark,
with consistent improvements in ISLR regardless of the underlying prediction
model architecture. This work has the potential to accelerate linguistic
research in the domain of signed languages and reduce communication barriers
between deaf and hearing people. | cs.CL | Improving Sign Recognition with Phonology |
2023-02-11T16:26:57Z | http://arxiv.org/pdf/2302.05737v2 | http://arxiv.org/abs/2302.05737v2 | Lin Zheng, Jianbo Yuan, Lei Yu, Lingpeng Kong | 2024-02-03T07:59:26Z | cs.CL, cs.LG | This work studies discrete diffusion probabilistic models with applications
to natural language generation. We derive an alternative yet equivalent
formulation of the sampling from discrete diffusion processes and leverage this
insight to develop a family of reparameterized discrete diffusion models. The
derived generic framework is highly flexible, offers a fresh perspective of the
generation process in discrete diffusion models, and features more effective
training and decoding techniques. We conduct extensive experiments to evaluate
the text generation capability of our model, demonstrating significant
improvements over existing diffusion models. | cs.CL | A Reparameterized Discrete Diffusion Model for Text Generation |
2023-02-11T16:04:38Z | http://arxiv.org/pdf/2302.05735v2 | http://arxiv.org/abs/2302.05735v2 | Alexander Pugantsov, Richard McCreadie | 2023-02-28T11:26:32Z | cs.CL | Transferring learned patterns from pretrained neural language models has been
shown to significantly improve effectiveness across a variety of language-based
tasks, meanwhile further tuning on intermediate tasks has been demonstrated to
provide additional performance benefits, provided the intermediate task is
sufficiently related to the target task. However, how to identify related tasks
is an open problem, and brute-force searching effective task combinations is
prohibitively expensive. Hence, the question arises, are we able to improve the
effectiveness and efficiency of tasks with no training examples through
selective fine-tuning? In this paper, we explore statistical measures that
approximate the divergence between domain representations as a means to
estimate whether tuning using one task pair will exhibit performance benefits
over tuning another. This estimation can then be used to reduce the number of
task pairs that need to be tested by eliminating pairs that are unlikely to
provide benefits. Through experimentation over 58 tasks and over 6,600 task
pair combinations, we demonstrate that statistical measures can distinguish
effective task pairs, and the resulting estimates can reduce end-to-end runtime
by up to 40%. | cs.CL | Divergence-Based Domain Transferability for Zero-Shot Classification |
2023-02-11T15:50:20Z | http://arxiv.org/pdf/2302.05729v2 | http://arxiv.org/abs/2302.05729v2 | Ha-Thanh Nguyen | 2023-02-14T06:26:42Z | cs.CL, cs.AI | LawGPT 1.0 is a virtual legal assistant built on the state-of-the-art
language model GPT-3, fine-tuned for the legal domain. The system is designed
to provide legal assistance to users in a conversational manner, helping them
with tasks such as answering legal questions, generating legal documents, and
providing legal advice. In this paper, we provide a brief overview of LawGPT
1.0, its architecture, and its performance on a set of legal benchmark tasks.
Please note that the detailed information about the model is protected by a
non-disclosure agreement (NDA) and cannot be disclosed in this report. | cs.CL | A Brief Report on LawGPT 1.0: A Virtual Legal Assistant Based on GPT-3 |
2023-02-11T15:34:23Z | http://arxiv.org/pdf/2302.05721v1 | http://arxiv.org/abs/2302.05721v1 | Varun Khurana, Yaman Kumar Singla, Nora Hollenstein, Rajesh Kumar, Balaji Krishnamurthy | 2023-02-11T15:34:23Z | cs.HC, cs.CL | Integrating human feedback in models can improve the performance of natural
language processing (NLP) models. Feedback can be either explicit (e.g. ranking
used in training language models) or implicit (e.g. using human cognitive
signals in the form of eyetracking). Prior eye tracking and NLP research reveal
that cognitive processes, such as human scanpaths, gleaned from human gaze
patterns aid in the understanding and performance of NLP models. However, the
collection of real eyetracking data for NLP tasks is challenging due to the
requirement of expensive and precise equipment coupled with privacy invasion
issues. To address this challenge, we propose ScanTextGAN, a novel model for
generating human scanpaths over text. We show that ScanTextGAN-generated
scanpaths can approximate meaningful cognitive signals in human gaze patterns.
We include synthetically generated scanpaths in four popular NLP tasks spanning
six different datasets as proof of concept and show that the models augmented
with generated scanpaths improve the performance of all downstream NLP tasks. | cs.HC | Synthesizing Human Gaze Feedback for Improved NLP Performance |
2023-02-11T15:15:41Z | http://arxiv.org/pdf/2302.05717v1 | http://arxiv.org/abs/2302.05717v1 | Jiayu Liu, Zhenya Huang, Chengxiang Zhai, Qi Liu | 2023-02-11T15:15:41Z | cs.AI, cs.CL, cs.LG | Mathematical reasoning is one of the crucial abilities of general artificial
intelligence, which requires machines to master mathematical logic and
knowledge from solving problems. However, existing approaches are not
transparent (thus not interpretable) in terms of what knowledge has been
learned and applied in the reasoning process. In this paper, we propose a
general Learning by Applying (LeAp) framework to enhance existing models
(backbones) in a principled way by explicit knowledge learning. In LeAp, we
perform knowledge learning in a novel problem-knowledge-expression paradigm,
with a Knowledge Encoder to acquire knowledge from problem data and a Knowledge
Decoder to apply knowledge for expression reasoning. The learned mathematical
knowledge, including word-word relations and word-operator relations, forms an
explicit knowledge graph, which bridges the knowledge "learning" and "applying"
organically. Moreover, for problem solving, we design a semantics-enhanced
module and a reasoning-enhanced module that apply knowledge to improve the
problem comprehension and symbol reasoning abilities of any backbone,
respectively. We theoretically prove the superiority of LeAp's autonomous
learning mechanism. Experiments on three real-world datasets show that LeAp
improves all backbones' performances, learns accurate knowledge, and achieves a
more interpretable reasoning process. | cs.AI | Learning by Applying: A General Framework for Mathematical Reasoning via Enhancing Explicit Knowledge Learning |
2023-02-11T14:54:00Z | http://arxiv.org/pdf/2302.05711v1 | http://arxiv.org/abs/2302.05711v1 | Xudong Han, Timothy Baldwin, Trevor Cohn | 2023-02-11T14:54:00Z | cs.CL, cs.CY, cs.LG | Modern NLP systems exhibit a range of biases, which a growing literature on
model debiasing attempts to correct. However current progress is hampered by a
plurality of definitions of bias, means of quantification, and oftentimes vague
relation between debiasing algorithms and theoretical measures of bias. This
paper seeks to clarify the current situation and plot a course for meaningful
progress in fair learning, with two key contributions: (1) making clear
inter-relations among the current gamut of methods, and their relation to
fairness theory; and (2) addressing the practical problem of model selection,
which involves a trade-off between fairness and accuracy and has led to
systemic issues in fairness research. Putting them together, we make several
recommendations to help shape future work. | cs.CL | Fair Enough: Standardizing Evaluation and Model Selection for Fairness Research in NLP |
2023-02-11T14:44:39Z | http://arxiv.org/pdf/2302.05706v1 | http://arxiv.org/abs/2302.05706v1 | Wenxuan Wang, Jen-tse Huang, Weibin Wu, Jianping Zhang, Yizhan Huang, Shuqing Li, Pinjia He, Michael Lyu | 2023-02-11T14:44:39Z | cs.CL, cs.SE | The exponential growth of social media platforms such as Twitter and Facebook
has revolutionized textual communication and textual content publication in
human society. However, they have been increasingly exploited to propagate
toxic content, such as hate speech, malicious advertisement, and pornography,
which can lead to highly negative impacts (e.g., harmful effects on teen mental
health). Researchers and practitioners have been enthusiastically developing
and extensively deploying textual content moderation software to address this
problem. However, we find that malicious users can evade moderation by changing
only a few words in the toxic content. Moreover, modern content moderation
software performance against malicious inputs remains underexplored. To this
end, we propose MTTM, a Metamorphic Testing framework for Textual content
Moderation software. Specifically, we conduct a pilot study on 2,000 text
messages collected from real users and summarize eleven metamorphic relations
across three perturbation levels: character, word, and sentence. MTTM employs
these metamorphic relations on toxic textual contents to generate test cases,
which are still toxic yet likely to evade moderation. In our evaluation, we
employ MTTM to test three commercial textual content moderation software and
two state-of-the-art moderation algorithms against three kinds of toxic
content. The results show that MTTM achieves up to 83.9%, 51%, and 82.5% error
finding rates (EFR) when testing commercial moderation software provided by
Google, Baidu, and Huawei, respectively, and it obtains up to 91.2% EFR when
testing the state-of-the-art algorithms from the academy. In addition, we
leverage the test cases generated by MTTM to retrain the model we explored,
which largely improves model robustness (0% to 5.9% EFR) while maintaining the
accuracy on the original test set. | cs.CL | MTTM: Metamorphic Testing for Textual Content Moderation Software |
2023-02-11T14:36:11Z | http://arxiv.org/pdf/2302.05703v1 | http://arxiv.org/abs/2302.05703v1 | Piush Aggarwal, Pranit Chawla, Mithun Das, Punyajoy Saha, Binny Mathew, Torsten Zesch, Animesh Mukherjee | 2023-02-11T14:36:11Z | cs.CL, cs.AI | Exploiting social media to spread hate has tremendously increased over the
years. Lately, multi-modal hateful content such as memes has drawn relatively
more traction than uni-modal content. Moreover, the availability of implicit
content payloads makes them fairly challenging to be detected by existing
hateful meme detection systems. In this paper, we present a use case study to
analyze such systems' vulnerabilities against external adversarial attacks. We
find that even very simple perturbations in uni-modal and multi-modal settings
performed by humans with little knowledge about the model can make the existing
detection models highly vulnerable. Empirically, we find a noticeable
performance drop of as high as 10% in the macro-F1 score for certain attacks.
As a remedy, we attempt to boost the model's robustness using contrastive
learning as well as an adversarial training-based method - VILLA. Using an
ensemble of the above two approaches, in two of our high resolution datasets,
we are able to (re)gain back the performance to a large extent for certain
attacks. We believe that ours is a first step toward addressing this crucial
problem in an adversarial setting and would inspire more such investigations in
the future. | cs.CL | HateProof: Are Hateful Meme Detection Systems really Robust? |
2023-02-11T14:02:08Z | http://arxiv.org/pdf/2302.05698v3 | http://arxiv.org/abs/2302.05698v3 | Jiacheng Ye, Zhiyong Wu, Jiangtao Feng, Tao Yu, Lingpeng Kong | 2023-06-20T08:37:42Z | cs.CL, cs.AI, cs.LG | Large pretrained language models (LMs) have shown impressive In-Context
Learning (ICL) ability, where the model learns to do an unseen task via a
prompt consisting of input-output examples as the demonstration, without any
parameter updates. The performance of ICL is highly dominated by the quality of
the selected in-context examples. However, previous selection methods are
mostly based on simple heuristics, leading to sub-optimal performance. In this
work, we formulate in-context example selection as a subset selection problem.
We propose CEIL (Compositional Exemplars for In-context Learning), which is
instantiated by Determinantal Point Processes (DPPs) to model the interaction
between the given input and in-context examples, and optimized through a
carefully-designed contrastive learning objective to obtain preference from
LMs. We validate CEIL on 12 classification and generation datasets from 7
distinct NLP tasks, including sentiment analysis, paraphrase detection, natural
language inference, commonsense reasoning, open-domain question answering, code
generation, and semantic parsing. Extensive experiments demonstrate not only
the state-of-the-art performance but also the transferability and
compositionality of CEIL, shedding new light on effective and efficient
in-context learning. Our code is released at
https://github.com/HKUNLP/icl-ceil. | cs.CL | Compositional Exemplars for In-context Learning |
2023-02-11T12:11:03Z | http://arxiv.org/pdf/2302.05674v1 | http://arxiv.org/abs/2302.05674v1 | Zhongbin Xie, Vid Kocijan, Thomas Lukasiewicz, Oana-Maria Camburu | 2023-02-11T12:11:03Z | cs.CL, cs.AI | Bias-measuring datasets play a critical role in detecting biased behavior of
language models and in evaluating progress of bias mitigation methods. In this
work, we focus on evaluating gender bias through coreference resolution, where
previous datasets are either hand-crafted or fail to reliably measure an
explicitly defined bias. To overcome these shortcomings, we propose a novel
method to collect diverse, natural, and minimally distant text pairs via
counterfactual generation, and construct Counter-GAP, an annotated dataset
consisting of 4008 instances grouped into 1002 quadruples. We further identify
a bias cancellation problem in previous group-level metrics on Counter-GAP, and
propose to use the difference between inconsistency across genders and within
genders to measure bias at a quadruple level. Our results show that four
pre-trained language models are significantly more inconsistent across
different gender groups than within each group, and that a name-based
counterfactual data augmentation method is more effective to mitigate such bias
than an anonymization-based method. | cs.CL | Counter-GAP: Counterfactual Bias Evaluation through Gendered Ambiguous Pronouns |
2023-02-11T11:42:39Z | http://arxiv.org/pdf/2302.07733v1 | http://arxiv.org/abs/2302.07733v1 | Yi Cai, Arthur Zimek, Eirini Ntoutsi, Gerhard Wunder | 2023-02-11T11:42:39Z | cs.CL, cs.LG | The importance of neighborhood construction in local explanation methods has
been already highlighted in the literature. And several attempts have been made
to improve neighborhood quality for high-dimensional data, for example, texts,
by adopting generative models. Although the generators produce more realistic
samples, the intuitive sampling approaches in the existing solutions leave the
latent space underexplored. To overcome this problem, our work, focusing on
local model-agnostic explanations for text classifiers, proposes a progressive
approximation approach that refines the neighborhood of a to-be-explained
decision with a careful two-stage interpolation using counterfactuals as
landmarks. We explicitly specify the two properties that should be satisfied by
generative models, the reconstruction ability and the locality-preserving
property, to guide the selection of generators for local explanation methods.
Moreover, noticing the opacity of generative models during the study, we
propose another method that implements progressive neighborhood approximation
with probability-based editions as an alternative to the generator-based
solution. The explanation results from both methods consist of word-level and
instance-level explanations benefiting from the realistic neighborhood. Through
exhaustive experiments, we qualitatively and quantitatively demonstrate the
effectiveness of the two proposed methods. | cs.CL | Explaining text classifiers through progressive neighborhood approximation with realistic samples |
2023-02-11T11:32:10Z | http://arxiv.org/pdf/2302.05658v2 | http://arxiv.org/abs/2302.05658v2 | Štěpán Šimsa, Milan Šulc, Michal Uřičář, Yash Patel, Ahmed Hamdi, Matěj Kocián, Matyáš Skalický, Jiří Matas, Antoine Doucet, Mickaël Coustaty, Dimosthenis Karatzas | 2023-05-03T16:24:58Z | cs.CL, cs.AI, cs.LG | This paper introduces the DocILE benchmark with the largest dataset of
business documents for the tasks of Key Information Localization and Extraction
and Line Item Recognition. It contains 6.7k annotated business documents, 100k
synthetically generated documents, and nearly~1M unlabeled documents for
unsupervised pre-training. The dataset has been built with knowledge of domain-
and task-specific aspects, resulting in the following key features: (i)
annotations in 55 classes, which surpasses the granularity of previously
published key information extraction datasets by a large margin; (ii) Line Item
Recognition represents a highly practical information extraction task, where
key information has to be assigned to items in a table; (iii) documents come
from numerous layouts and the test set includes zero- and few-shot cases as
well as layouts commonly seen in the training set. The benchmark comes with
several baselines, including RoBERTa, LayoutLMv3 and DETR-based Table
Transformer; applied to both tasks of the DocILE benchmark, with results shared
in this paper, offering a quick starting point for future work. The dataset,
baselines and supplementary material are available at
https://github.com/rossumai/docile. | cs.CL | DocILE Benchmark for Document Information Localization and Extraction |
2023-02-11T11:32:08Z | http://arxiv.org/pdf/2302.05657v1 | http://arxiv.org/abs/2302.05657v1 | Thyge Enggaard, August Lohse, Morten Axel Pedersen, Sune Lehmann | 2023-02-11T11:32:08Z | cs.CL | Word embeddings provide an unsupervised way to understand differences in word
usage between discursive communities. A number of recent papers have focused on
identifying words that are used differently by two or more communities. But
word embeddings are complex, high-dimensional spaces and a focus on identifying
differences only captures a fraction of their richness. Here, we take a step
towards leveraging the richness of the full embedding space, by using word
embeddings to map out how words are used differently. Specifically, we describe
the construction of dialectograms, an unsupervised way to visually explore the
characteristic ways in which each community use a focal word. Based on these
dialectograms, we provide a new measure of the degree to which words are used
differently that overcomes the tendency for existing measures to pick out low
frequent or polysemous words. We apply our methods to explore the discourses of
two US political subreddits and show how our methods identify stark affective
polarisation of politicians and political entities, differences in the
assessment of proper political action as well as disagreement about whether
certain issues require political intervention at all. | cs.CL | Dialectograms: Machine Learning Differences between Discursive Communities |
2023-02-11T07:30:21Z | http://arxiv.org/pdf/2302.10276v2 | http://arxiv.org/abs/2302.10276v2 | Likun Yang, Xiaokun Feng, Xiaotang Chen, Shiyu Zhang, Kaiqi Huang | 2023-03-16T10:01:05Z | cs.CV, cs.AI, cs.CL | In psychoanalysis, generating interpretations to one's psychological state
through visual creations is facing significant demands. The two main tasks of
existing studies in the field of computer vision, sentiment/emotion
classification and affective captioning, can hardly satisfy the requirement of
psychological interpreting. To meet the demands for psychoanalysis, we
introduce a challenging task, \textbf{V}isual \textbf{E}motion
\textbf{I}nterpretation \textbf{T}ask (VEIT). VEIT requires AI to generate
reasonable interpretations of creator's psychological state through visual
creations. To support the task, we present a multimodal dataset termed SpyIn
(\textbf{S}and\textbf{p}la\textbf{y} \textbf{In}terpretation Dataset), which is
psychological theory supported and professional annotated. Dataset analysis
illustrates that SpyIn is not only able to support VEIT, but also more
challenging compared with other captioning datasets. Building on SpyIn, we
conduct experiments of several image captioning method, and propose a
visual-semantic combined model which obtains a SOTA result on SpyIn. The
results indicate that VEIT is a more challenging task requiring scene graph
information and psychological knowledge. Our work also show a promise for AI to
analyze and explain inner world of humanity through visual creations. | cs.CV | See Your Heart: Psychological states Interpretation through Visual Creations |
2023-02-11T07:01:53Z | http://arxiv.org/pdf/2302.05619v1 | http://arxiv.org/abs/2302.05619v1 | Yoichi Ishibashi, Danushka Bollegala, Katsuhito Sudoh, Satoshi Nakamura | 2023-02-11T07:01:53Z | cs.CL, cs.AI | Discrete prompts have been used for fine-tuning Pre-trained Language Models
for diverse NLP tasks. In particular, automatic methods that generate discrete
prompts from a small set of training instances have reported superior
performance. However, a closer look at the learnt prompts reveals that they
contain noisy and counter-intuitive lexical constructs that would not be
encountered in manually-written prompts. This raises an important yet
understudied question regarding the robustness of automatically learnt discrete
prompts when used in downstream tasks. To address this question, we conduct a
systematic study of the robustness of discrete prompts by applying carefully
designed perturbations into an application using AutoPrompt and then measure
their performance in two Natural Language Inference (NLI) datasets. Our
experimental results show that although the discrete prompt-based method
remains relatively robust against perturbations to NLI inputs, they are highly
sensitive to other types of perturbations such as shuffling and deletion of
prompt tokens. Moreover, they generalize poorly across different NLI datasets.
We hope our findings will inspire future work on robust discrete prompt
learning. | cs.CL | Evaluating the Robustness of Discrete Prompts |
2023-02-11T05:53:51Z | http://arxiv.org/pdf/2302.05611v1 | http://arxiv.org/abs/2302.05611v1 | Shun Wang, Yucheng Li, Chenghua Lin, Loïc Barrault, Frank Guerin | 2023-02-11T05:53:51Z | cs.CL | We propose a novel RoBERTa-based model, RoPPT, which introduces a
target-oriented parse tree structure in metaphor detection. Compared to
existing models, RoPPT focuses on semantically relevant information and
achieves the state-of-the-art on several main metaphor datasets. We also
compare our approach against several popular denoising and pruning methods,
demonstrating the effectiveness of our approach in context denoising. Our code
and dataset can be found at https://github.com/MajiBear000/RoPPT | cs.CL | Metaphor Detection with Effective Context Denoising |
2023-02-11T05:52:33Z | http://arxiv.org/pdf/2302.05610v1 | http://arxiv.org/abs/2302.05610v1 | Md Mahbubur Rahman, Shaila Shova | 2023-02-11T05:52:33Z | cs.LG, cs.AI, cs.CL | Over the last few years, social media has evolved into a medium for
expressing personal views, emotions, and even business and political proposals,
recommendations, and advertisements. We address the topic of identifying
emotions from text data obtained from social media posts like Twitter in this
research. We have deployed different traditional machine learning techniques
such as Support Vector Machines (SVM), Naive Bayes, Decision Trees, and Random
Forest, as well as deep neural network models such as LSTM, CNN, GRU, BiLSTM,
BiGRU to classify these tweets into four emotion categories (Fear, Anger, Joy,
and Sadness). Furthermore, we have constructed a BiLSTM and BiGRU ensemble
model. The evaluation result shows that the deep neural network models(BiGRU,
to be specific) produce the most promising results compared to traditional
machine learning models, with an 87.53 % accuracy rate. The ensemble model
performs even better (87.66 %), albeit the difference is not significant. This
result will aid in the development of a decision-making tool that visualizes
emotional fluctuations. | cs.LG | Emotion Detection From Social Media Posts |
2023-02-11T05:46:21Z | http://arxiv.org/pdf/2302.05608v1 | http://arxiv.org/abs/2302.05608v1 | Zhu Wang, Sourav Medya, Sathya N. Ravi | 2023-02-11T05:46:21Z | cs.CV, cs.AI, cs.CL, cs.LG | Often, deep network models are purely inductive during training and while
performing inference on unseen data. Thus, when such models are used for
predictions, it is well known that they often fail to capture the semantic
information and implicit dependencies that exist among objects (or concepts) on
a population level. Moreover, it is still unclear how domain or prior modal
knowledge can be specified in a backpropagation friendly manner, especially in
large-scale and noisy settings. In this work, we propose an end-to-end vision
and language model incorporating explicit knowledge graphs. We also introduce
an interactive out-of-distribution (OOD) layer using implicit network operator.
The layer is used to filter noise that is brought by external knowledge base.
In practice, we apply our model on several vision and language downstream tasks
including visual question answering, visual reasoning, and image-text retrieval
on different datasets. Our experiments show that it is possible to design
models that perform similarly to state-of-art results but with significantly
fewer samples and training time. | cs.CV | Differentiable Outlier Detection Enable Robust Deep Multimodal Analysis |
2023-02-11T04:18:07Z | http://arxiv.org/pdf/2302.05597v1 | http://arxiv.org/abs/2302.05597v1 | Xianjun Yang, Stephen Wilson, Linda Petzold | 2023-02-11T04:18:07Z | cs.CL, cs.AI | In this paper, we present a novel approach to knowledge extraction and
retrieval using Natural Language Processing (NLP) techniques for material
science. Our goal is to automatically mine structured knowledge from millions
of research articles in the field of polycrystalline materials and make it
easily accessible to the broader community. The proposed method leverages NLP
techniques such as entity recognition and document classification to extract
relevant information and build an extensive knowledge base, from a collection
of 9.5 Million publications. The resulting knowledge base is integrated into a
search engine, which enables users to search for information about specific
materials, properties, and experiments with greater precision than traditional
search engines like Google. We hope our results can enable material scientists
quickly locate desired experimental procedures, compare their differences, and
even inspire them to design new experiments. Our website will be available at
Github \footnote{https://github.com/Xianjun-Yang/PcMSP.git} soon. | cs.CL | MatKB: Semantic Search for Polycrystalline Materials Synthesis Procedures |
2023-02-11T03:13:54Z | http://arxiv.org/pdf/2302.07736v2 | http://arxiv.org/abs/2302.07736v2 | Fan Huang, Haewoon Kwak, Jisun An | 2023-03-15T19:16:45Z | cs.CL, cs.HC | Recent studies have alarmed that many online hate speeches are implicit. With
its subtle nature, the explainability of the detection of such hateful speech
has been a challenging problem. In this work, we examine whether ChatGPT can be
used for providing natural language explanations (NLEs) for implicit hateful
speech detection. We design our prompt to elicit concise ChatGPT-generated NLEs
and conduct user studies to evaluate their qualities by comparison with
human-written NLEs. We discuss the potential and limitations of ChatGPT in the
context of implicit hateful speech research. | cs.CL | Is ChatGPT better than Human Annotators? Potential and Limitations of ChatGPT in Explaining Implicit Hate Speech |
2023-02-11T02:53:12Z | http://arxiv.org/pdf/2302.05582v1 | http://arxiv.org/abs/2302.05582v1 | Daniel Hao Xian Yuen, Andrew Yong Chen Pang, Zhou Yang, Chun Yong Chong, Mei Kuan Lim, David Lo | 2023-02-11T02:53:12Z | eess.AS, cs.CL, cs.SD, cs.SE | Recent years have witnessed wider adoption of Automated Speech Recognition
(ASR) techniques in various domains. Consequently, evaluating and enhancing the
quality of ASR systems is of great importance. This paper proposes ASDF, an
Automated Speech Recognition Differential Testing Framework for testing ASR
systems. ASDF extends an existing ASR testing tool, the CrossASR++, which
synthesizes test cases from a text corpus. However, CrossASR++ fails to make
use of the text corpus efficiently and provides limited information on how the
failed test cases can improve ASR systems. To address these limitations, our
tool incorporates two novel features: (1) a text transformation module to boost
the number of generated test cases and uncover more errors in ASR systems and
(2) a phonetic analysis module to identify on which phonemes the ASR system
tend to produce errors. ASDF generates more high-quality test cases by applying
various text transformation methods (e.g., change tense) to the texts in failed
test cases. By doing so, ASDF can utilize a small text corpus to generate a
large number of audio test cases, something which CrossASR++ is not capable of.
In addition, ASDF implements more metrics to evaluate the performance of ASR
systems from multiple perspectives. ASDF performs phonetic analysis on the
identified failed test cases to identify the phonemes that ASR systems tend to
transcribe incorrectly, providing useful information for developers to improve
ASR systems. The demonstration video of our tool is made online at
https://www.youtube.com/watch?v=DzVwfc3h9As. The implementation is available at
https://github.com/danielyuenhx/asdf-differential-testing. | eess.AS | ASDF: A Differential Testing Framework for Automatic Speech Recognition Systems |
2023-02-11T02:43:34Z | http://arxiv.org/pdf/2302.05578v2 | http://arxiv.org/abs/2302.05578v2 | Renat Aksitov, Chung-Ching Chang, David Reitter, Siamak Shakeri, Yunhsuan Sung | 2023-02-14T23:53:56Z | cs.CL, cs.AI | Despite recent progress, it has been difficult to prevent semantic
hallucinations in generative Large Language Models. One common solution to this
is augmenting LLMs with a retrieval system and making sure that the generated
output is attributable to the retrieved information. Given this new added
constraint, it is plausible to expect that the overall quality of the output
will be affected, for example, in terms of fluency. Can scaling language models
help?
Here we examine the relationship between fluency and attribution in LLMs
prompted with retrieved evidence in knowledge-heavy dialog settings. Our
experiments were implemented with a set of auto-metrics that are aligned with
human preferences. They were used to evaluate a large set of generations,
produced under varying parameters of LLMs and supplied context.
We show that larger models tend to do much better in both fluency and
attribution, and that (naively) using top-k retrieval versus top-1 retrieval
improves attribution but hurts fluency. We next propose a recipe that could
allow smaller models to both close the gap with larger models and preserve the
benefits of top-k retrieval while avoiding its drawbacks. | cs.CL | Characterizing Attribution and Fluency Tradeoffs for Retrieval-Augmented Large Language Models |
2023-02-11T02:20:25Z | http://arxiv.org/pdf/2302.05574v1 | http://arxiv.org/abs/2302.05574v1 | Junru Lu, Jiazheng Li, Byron C. Wallace, Yulan He, Gabriele Pergola | 2023-02-11T02:20:25Z | cs.CL | Accessing medical literature is difficult for laypeople as the content is
written for specialists and contains medical jargon. Automated text
simplification methods offer a potential means to address this issue. In this
work, we propose a summarize-then-simplify two-stage strategy, which we call
NapSS, identifying the relevant content to simplify while ensuring that the
original narrative flow is preserved. In this approach, we first generate
reference summaries via sentence matching between the original and the
simplified abstracts. These summaries are then used to train an extractive
summarizer, learning the most relevant content to be simplified. Then, to
ensure the narrative consistency of the simplified text, we synthesize
auxiliary narrative prompts combining key phrases derived from the syntactical
analyses of the original text. Our model achieves results significantly better
than the seq2seq baseline on an English medical corpus, yielding 3%~4% absolute
improvements in terms of lexical similarity, and providing a further 1.1%
improvement of SARI score when combined with the baseline. We also highlight
shortcomings of existing evaluation methods, and introduce new metrics that
take into account both lexical and high-level semantic similarity. A human
evaluation conducted on a random sample of the test set further establishes the
effectiveness of the proposed approach. Codes and models are released here:
https://github.com/LuJunru/NapSS. | cs.CL | NapSS: Paragraph-level Medical Text Simplification via Narrative Prompting and Sentence-matching Summarization |
2023-02-10T20:54:10Z | http://arxiv.org/pdf/2302.05508v1 | http://arxiv.org/abs/2302.05508v1 | Hrishikesh Viswanath, Tianyi Zhang | 2023-02-10T20:54:10Z | cs.CL, cs.AI, cs.LG | Studies have shown that large pretrained language models exhibit biases
against social groups based on race, gender etc, which they inherit from the
datasets they are trained on. Various researchers have proposed mathematical
tools for quantifying and identifying these biases. There have been methods
proposed to mitigate such biases. In this paper, we present a comprehensive
quantitative evaluation of different kinds of biases such as race, gender,
ethnicity, age etc. exhibited by popular pretrained language models such as
BERT, GPT-2 etc. and also present a toolkit that provides plug-and-play
interfaces to connect mathematical tools to identify biases with large
pretrained language models such as BERT, GPT-2 etc. and also present users with
the opportunity to test custom models against these metrics. The toolkit also
allows users to debias existing and custom models using the debiasing
techniques proposed so far. The toolkit is available at
https://github.com/HrishikeshVish/Fairpy. | cs.CL | FairPy: A Toolkit for Evaluation of Social Biases and their Mitigation in Large Language Models |
2023-02-10T20:50:58Z | http://arxiv.org/pdf/2302.05507v2 | http://arxiv.org/abs/2302.05507v2 | Nicolas Gontier, Pau Rodriguez, Issam Laradji, David Vazquez, Christopher Pal | 2023-11-17T19:30:10Z | cs.CL, cs.AI, cs.LG | Text-based game environments are challenging because agents must deal with
long sequences of text, execute compositional actions using text and learn from
sparse rewards. We address these challenges by proposing Language Decision
Transformers (LDTs), a framework that is based on transformer language models
and decision transformers (DTs). Our LDTs extend DTs with 3 components: (1)
exponential tilt to guide the agent towards high obtainable goals, (2) novel
goal conditioning methods yielding better results than the traditional
return-to-go (sum of all future rewards), and (3) a model of future
observations that improves agent performance. LDTs are the first to address
offline RL with DTs on these challenging games. Our experiments show that LDTs
achieve the highest scores among many different types of agents on some of the
most challenging Jericho games, such as Enchanter. | cs.CL | Language Decision Transformers with Exponential Tilt for Interactive Text Environments |
2023-02-10T20:23:27Z | http://arxiv.org/pdf/2302.07738v2 | http://arxiv.org/abs/2302.07738v2 | Antoine Lefebvre-Brossard, Stephane Gazaille, Michel C. Desmarais | 2023-04-14T13:20:07Z | cs.CL, cs.IR, cs.LG | Teachers and students are increasingly relying on online learning resources
to supplement the ones provided in school. This increase in the breadth and
depth of available resources is a great thing for students, but only provided
they are able to find answers to their queries. Question-answering and
information retrieval systems have benefited from public datasets to train and
evaluate their algorithms, but most of these datasets have been in English text
written by and for adults. We introduce a new public French question-answering
dataset collected from Alloprof, a Quebec-based primary and high-school help
website, containing 29 349 questions and their explanations in a variety of
school subjects from 10 368 students, with more than half of the explanations
containing links to other questions or some of the 2 596 reference pages on the
website. We also present a case study of this dataset in an information
retrieval task. This dataset was collected on the Alloprof public forum, with
all questions verified for their appropriateness and the explanations verified
both for their appropriateness and their relevance to the question. To predict
relevant documents, architectures using pre-trained BERT models were fine-tuned
and evaluated. This dataset will allow researchers to develop
question-answering, information retrieval and other algorithms specifically for
the French speaking education context. Furthermore, the range of language
proficiency, images, mathematical symbols and spelling mistakes will
necessitate algorithms based on a multimodal comprehension. The case study we
present as a baseline shows an approach that relies on recent techniques
provides an acceptable performance level, but more work is necessary before it
can reliably be used and trusted in a production setting. | cs.CL | Alloprof: a new French question-answer education dataset and its use in an information retrieval case study |
2023-02-10T19:40:10Z | http://arxiv.org/pdf/2302.07731v3 | http://arxiv.org/abs/2302.07731v3 | Alessandro Gambetti, Qiwei Han | 2023-04-21T16:40:55Z | cs.CL, cs.AI, cs.IR, cs.LG | Recent advances in generative models such as GPT may be used to fabricate
indistinguishable fake customer reviews at a much lower cost, thus posing
challenges for social media platforms to detect these machine-generated fake
reviews. We propose to leverage the high-quality elite restaurant reviews
verified by Yelp to generate fake reviews from the OpenAI GPT review creator
and ultimately fine-tune a GPT output detector to predict fake reviews that
significantly outperform existing solutions. We further apply the model to
predict non-elite reviews and identify the patterns across several dimensions,
such as review, user and restaurant characteristics, and writing style. We show
that social media platforms are continuously challenged by machine-generated
fake reviews, although they may implement detection systems to filter out
suspicious reviews. | cs.CL | Combat AI With AI: Counteract Machine-Generated Fake Restaurant Reviews on Social Media |
2023-02-10T19:00:00Z | http://arxiv.org/pdf/2302.05454v1 | http://arxiv.org/abs/2302.05454v1 | Marco Farina, Duccio Pappadopulo, Anant Gupta, Leslie Huang, Ozan İrsoy, Thamar Solorio | 2023-02-10T19:00:00Z | cs.CL, cs.IR | Driven by encouraging results on a wide range of tasks, the field of NLP is
experiencing an accelerated race to develop bigger language models. This race
for bigger models has also underscored the need to continue the pursuit of
practical distillation approaches that can leverage the knowledge acquired by
these big models in a compute-efficient manner. Having this goal in mind, we
build on recent work to propose a hallucination-free framework for sequence
tagging that is especially suited for distillation. We show empirical results
of new state-of-the-art performance across multiple sequence labelling datasets
and validate the usefulness of this framework for distilling a large model in a
few-shot learning scenario. | cs.CL | Distillation of encoder-decoder transformers for sequence labelling |
2023-02-10T18:21:13Z | http://arxiv.org/pdf/2302.05406v1 | http://arxiv.org/abs/2302.05406v1 | Pedro Colon-Hernandez, Henry Lieberman, Yida Xin, Claire Yin, Cynthia Breazeal, Peter Chin | 2023-02-10T18:21:13Z | cs.CL | Contextualized or discourse aware commonsense inference is the task of
generating coherent commonsense assertions (i.e., facts) from a given story,
and a particular sentence from that story. Some problems with the task are:
lack of controllability for topics of the inferred facts; lack of commonsense
knowledge during training; and, possibly, hallucinated or false facts. In this
work, we utilize a transformer model for this task and develop techniques to
address the aforementioned problems in the task. We control the inference by
introducing a new technique we call "hinting". Hinting is a kind of language
model prompting, that utilizes both hard prompts (specific words) and soft
prompts (virtual learnable templates). This serves as a control signal to
advise the language model "what to talk about". Next, we establish a
methodology for performing joint inference with multiple commonsense knowledge
bases. Joint inference of commonsense requires care, because it is imprecise
and the level of generality is more flexible. You want to be sure that the
results "still make sense" for the context. To this end, we align the textual
version of assertions from three knowledge graphs (ConceptNet, ATOMIC2020, and
GLUCOSE) with a story and a target sentence. This combination allows us to
train a single model to perform joint inference with multiple knowledge graphs.
We show experimental results for the three knowledge graphs on joint inference.
Our final contribution is exploring a GAN architecture that generates the
contextualized commonsense assertions and scores them as to their plausibility
through a discriminator. The result is an integrated system for contextual
commonsense inference in stories, that can controllably generate plausible
commonsense assertions, and takes advantage of joint inference between multiple
commonsense knowledge bases. | cs.CL | Adversarial Transformer Language Models for Contextual Commonsense Inference |
2023-02-10T17:40:51Z | http://arxiv.org/pdf/2302.05392v1 | http://arxiv.org/abs/2302.05392v1 | Nhung T. H. Nguyen, Makoto Miwa, Sophia Ananiadou | 2023-02-10T17:40:51Z | cs.CL, cs.LG | The information bottleneck (IB) principle has been proven effective in
various NLP applications. The existing work, however, only used either
generative or information compression models to improve the performance of the
target task. In this paper, we propose to combine the two types of IB models
into one system to enhance Named Entity Recognition (NER). For one type of IB
model, we incorporate two unsupervised generative components, span
reconstruction and synonym generation, into a span-based NER system. The span
reconstruction ensures that the contextualised span representation keeps the
span information, while the synonym generation makes synonyms have similar
representations even in different contexts. For the other type of IB model, we
add a supervised IB layer that performs information compression into the system
to preserve useful features for NER in the resulting span representations.
Experiments on five different corpora indicate that jointly training both
generative and information compression models can enhance the performance of
the baseline span-based NER system. Our source code is publicly available at
https://github.com/nguyennth/joint-ib-models. | cs.CL | Span-based Named Entity Recognition by Generating and Compressing Information |
2023-02-10T13:49:50Z | http://arxiv.org/pdf/2302.05244v5 | http://arxiv.org/abs/2302.05244v5 | Laetitia Teodorescu, Xingdi Yuan, Marc-Alexandre Côté, Pierre-Yves Oudeyer | 2023-02-24T10:15:45Z | cs.AI, cs.CL, cs.LG | Building open-ended agents that can autonomously discover a diversity of
behaviours is one of the long-standing goals of artificial intelligence. This
challenge can be studied in the framework of autotelic RL agents, i.e. agents
that learn by selecting and pursuing their own goals, self-organizing a
learning curriculum. Recent work identified language as a key dimension of
autotelic learning, in particular because it enables abstract goal sampling and
guidance from social peers for hindsight relabelling. Within this perspective,
we study the following open scientific questions: What is the impact of
hindsight feedback from a social peer (e.g. selective vs. exhaustive)? How can
the agent learn from very rare language goal examples in its experience replay?
How can multiple forms of exploration be combined, and take advantage of easier
goals as stepping stones to reach harder ones? To address these questions, we
use ScienceWorld, a textual environment with rich abstract and combinatorial
physics. We show the importance of selectivity from the social peer's feedback;
that experience replay needs to over-sample examples of rare goals; and that
following self-generated goal sequences where the agent's competence is
intermediate leads to significant improvements in final performance. | cs.AI | A Song of Ice and Fire: Analyzing Textual Autotelic Agents in ScienceWorld |
2023-02-10T12:16:38Z | http://arxiv.org/pdf/2302.05206v1 | http://arxiv.org/abs/2302.05206v1 | Tianjun Zhang, Fangchen Liu, Justin Wong, Pieter Abbeel, Joseph E. Gonzalez | 2023-02-10T12:16:38Z | cs.CL, cs.AI | Reinforcement learning has seen wide success in finetuning large language
models to better align with instructions via human feedback. The so-called
algorithm, Reinforcement Learning with Human Feedback (RLHF) demonstrates
impressive performance on the GPT series models. However, the underlying
Reinforcement Learning (RL) algorithm is complex and requires an additional
training pipeline for reward and value networks. In this paper, we consider an
alternative approach: converting feedback to instruction by relabeling the
original one and training the model for better alignment in a supervised
manner. Such an algorithm doesn't require any additional parameters except for
the original language model and maximally reuses the pretraining pipeline. To
achieve this, we formulate instruction alignment problem for language models as
a goal-reaching problem in decision making. We propose Hindsight Instruction
Relabeling (HIR), a novel algorithm for aligning language models with
instructions. The resulting two-stage algorithm shed light to a family of
reward-free approaches that utilize the hindsightly relabeled instructions
based on feedback. We evaluate the performance of HIR extensively on 12
challenging BigBench reasoning tasks and show that HIR outperforms the baseline
algorithms and is comparable to or even surpasses supervised finetuning. | cs.CL | The Wisdom of Hindsight Makes Language Models Better Instruction Followers |
2023-02-10T09:43:15Z | http://arxiv.org/pdf/2302.05138v2 | http://arxiv.org/abs/2302.05138v2 | Liang Li, Ruiying Geng, Chengyang Fang, Bing Li, Can Ma, Binhua Li, Yongbin Li | 2023-02-28T07:20:06Z | cs.CL | Table-to-text generation aims at automatically generating text to help people
conveniently obtain salient information in tables. Recent works explicitly
decompose the generation process into content planning and surface generation
stages, employing two autoregressive networks for them respectively. However,
they are computationally expensive due to the non-parallelizable nature of
autoregressive decoding and the redundant parameters of two networks. In this
paper, we propose the first totally non-autoregressive table-to-text model
(Plan-then-Seam, PTS) that produces its outputs in parallel with one single
network. PTS firstly writes and calibrates one plan of the content to be
generated with a novel rethinking pointer predictor, and then takes the plan as
the context for seaming to decode the description. These two steps share
parameters and perform iteratively to capture token inter-dependency while
keeping parallel decoding. Experiments on two public benchmarks show that PTS
achieves 3.0~5.6 times speedup for inference time, reducing 50% parameters,
while maintaining as least comparable performance against strong two-stage
table-to-text competitors. | cs.CL | Plan-then-Seam: Towards Efficient Table-to-Text Generation |
2023-02-10T09:42:07Z | http://arxiv.org/pdf/2302.05137v1 | http://arxiv.org/abs/2302.05137v1 | Soyeong Jeong, Jinheon Baek, Sung Ju Hwang, Jong C. Park | 2023-02-10T09:42:07Z | cs.CL, cs.AI, cs.IR | Conversational Question Answering (ConvQA) models aim at answering a question
with its relevant paragraph and previous question-answer pairs that occurred
during conversation multiple times. To apply such models to a real-world
scenario, some existing work uses predicted answers, instead of unavailable
ground-truth answers, as the conversation history for inference. However, since
these models usually predict wrong answers, using all the predictions without
filtering significantly hampers the model performance. To address this problem,
we propose to filter out inaccurate answers in the conversation history based
on their estimated confidences and uncertainties from the ConvQA model, without
making any architectural changes. Moreover, to make the confidence and
uncertainty values more reliable, we propose to further calibrate them, thereby
smoothing the model predictions. We validate our models, Answer Selection-based
realistic Conversation Question Answering, on two standard ConvQA datasets, and
the results show that our models significantly outperform relevant baselines.
Code is available at: https://github.com/starsuzi/AS-ConvQA. | cs.CL | Realistic Conversational Question Answering with Answer Selection based on Calibrated Confidence and Uncertainty Measurement |
2023-02-10T09:17:52Z | http://arxiv.org/pdf/2302.05128v1 | http://arxiv.org/abs/2302.05128v1 | Yaqi Xie, Chen Yu, Tongyao Zhu, Jinbin Bai, Ze Gong, Harold Soh | 2023-02-10T09:17:52Z | cs.CL, cs.AI, cs.RO | Recent large language models (LLMs) have demonstrated remarkable performance
on a variety of natural language processing (NLP) tasks, leading to intense
excitement about their applicability across various domains. Unfortunately,
recent work has also shown that LLMs are unable to perform accurate reasoning
nor solve planning problems, which may limit their usefulness for
robotics-related tasks. In this work, our central question is whether LLMs are
able to translate goals specified in natural language to a structured planning
language. If so, LLM can act as a natural interface between the planner and
human users; the translated goal can be handed to domain-independent AI
planners that are very effective at planning. Our empirical results on GPT 3.5
variants show that LLMs are much better suited towards translation rather than
planning. We find that LLMs are able to leverage commonsense knowledge and
reasoning to furnish missing details from under-specified goals (as is often
the case in natural language). However, our experiments also reveal that LLMs
can fail to generate goals in tasks that involve numerical or physical (e.g.,
spatial) reasoning, and that LLMs are sensitive to the prompts used. As such,
these models are promising for translation to structured planning languages,
but care should be taken in their use. | cs.CL | Translating Natural Language to Planning Goals with Large-Language Models |
2023-02-10T08:50:51Z | http://arxiv.org/pdf/2302.05120v1 | http://arxiv.org/abs/2302.05120v1 | Piotr Gaiński, Klaudia Bałazy | 2023-02-10T08:50:51Z | cs.CL | We propose a novel gradient-based attack against transformer-based language
models that searches for an adversarial example in a continuous space of token
probabilities. Our algorithm mitigates the gap between adversarial loss for
continuous and discrete text representations by performing multi-step
quantization in a quantization-compensation loop. Experiments show that our
method significantly outperforms other approaches on various natural language
processing (NLP) tasks. | cs.CL | Step by Step Loss Goes Very Far: Multi-Step Quantization for Adversarial Text Attacks |
2023-02-10T08:21:42Z | http://arxiv.org/pdf/2302.05110v1 | http://arxiv.org/abs/2302.05110v1 | Spandan Dey, Md Sahidullah, Goutam Saha | 2023-02-10T08:21:42Z | eess.AS, cs.CL, cs.LG, cs.SD | This work addresses the cross-corpora generalization issue for the
low-resourced spoken language identification (LID) problem. We have conducted
the experiments in the context of Indian LID and identified strikingly poor
cross-corpora generalization due to corpora-dependent non-lingual biases. Our
contribution to this work is twofold. First, we propose domain diversification,
which diversifies the limited training data using different audio data
augmentation methods. We then propose the concept of maximally diversity-aware
cascaded augmentations and optimize the augmentation fold-factor for effective
diversification of the training data. Second, we introduce the idea of domain
generalization considering the augmentation methods as pseudo-domains. Towards
this, we investigate both domain-invariant and domain-aware approaches. Our LID
system is based on the state-of-the-art emphasized channel attention,
propagation, and aggregation based time delay neural network (ECAPA-TDNN)
architecture. We have conducted extensive experiments with three widely used
corpora for Indian LID research. In addition, we conduct a final blind
evaluation of our proposed methods on the Indian subset of VoxLingua107 corpus
collected in the wild. Our experiments demonstrate that the proposed domain
diversification is more promising over commonly used simple augmentation
methods. The study also reveals that domain generalization is a more effective
solution than domain diversification. We also notice that domain-aware learning
performs better for same-corpora LID, whereas domain-invariant learning is more
suitable for cross-corpora generalization. Compared to basic ECAPA-TDNN, its
proposed domain-invariant extensions improve the cross-corpora EER up to 5.23%.
In contrast, the proposed domain-aware extensions also improve performance for
same-corpora test scenarios. | eess.AS | Cross-Corpora Spoken Language Identification with Domain Diversification and Generalization |
2023-02-10T07:37:49Z | http://arxiv.org/pdf/2302.05096v1 | http://arxiv.org/abs/2302.05096v1 | Yen-Ting Lin, Alexandros Papangelis, Seokhwan Kim, Sungjin Lee, Devamanyu Hazarika, Mahdi Namazifar, Di Jin, Yang Liu, Dilek Hakkani-Tur | 2023-02-10T07:37:49Z | cs.CL, cs.AI | This work focuses on in-context data augmentation for intent detection.
Having found that augmentation via in-context prompting of large pre-trained
language models (PLMs) alone does not improve performance, we introduce a novel
approach based on PLMs and pointwise V-information (PVI), a metric that can
measure the usefulness of a datapoint for training a model. Our method first
fine-tunes a PLM on a small seed of training data and then synthesizes new
datapoints - utterances that correspond to given intents. It then employs
intent-aware filtering, based on PVI, to remove datapoints that are not helpful
to the downstream intent classifier. Our method is thus able to leverage the
expressive power of large language models to produce diverse training data.
Empirical results demonstrate that our method can produce synthetic training
data that achieve state-of-the-art performance on three challenging intent
detection datasets under few-shot settings (1.28% absolute improvement in
5-shot and 1.18% absolute in 10-shot, on average) and perform on par with the
state-of-the-art in full-shot settings (within 0.01% absolute, on average). | cs.CL | Selective In-Context Data Augmentation for Intent Detection using Pointwise V-Information |
2023-02-10T05:39:29Z | http://arxiv.org/pdf/2302.05061v1 | http://arxiv.org/abs/2302.05061v1 | Zhen Wang, Peide Zhu, Jie Yang | 2023-02-10T05:39:29Z | cs.CL | Controversy is widespread online. Previous studies mainly define controversy
based on vague assumptions of its relation to sentiment such as hate speech and
offensive words. This paper introduces the first question-answering dataset
that defines content controversy by user perception, i.e., votes from plenty of
users. It contains nearly 10K questions, and each question has a best answer
and a most controversial answer. Experimental results reveal that controversy
detection in question answering is essential and challenging, and there is no
strong correlation between controversy and sentiment tasks. | cs.CL | ControversialQA: Exploring Controversy in Question Answering |
2023-02-10T04:05:24Z | http://arxiv.org/pdf/2302.05040v2 | http://arxiv.org/abs/2302.05040v2 | Ziji Zhang, Zhehui Wang, Rajesh Kamma, Sharanya Eswaran, Narayanan Sadagopan | 2023-06-21T17:44:58Z | cs.CL, cs.LG, cs.SD, eess.AS | Speech-to-text errors made by automatic speech recognition (ASR) systems
negatively impact downstream models. Error correction models as a
post-processing text editing method have been recently developed for refining
the ASR outputs. However, efficient models that meet the low latency
requirements of industrial grade production systems have not been well studied.
We propose PATCorrect-a novel non-autoregressive (NAR) approach based on
multi-modal fusion leveraging representations from both text and phoneme
modalities, to reduce word error rate (WER) and perform robustly with varying
input transcription quality. We demonstrate that PATCorrect consistently
outperforms state-of-the-art NAR method on English corpus across different
upstream ASR systems, with an overall 11.62% WER reduction (WERR) compared to
9.46% WERR achieved by other methods using text only modality. Besides, its
inference latency is at tens of milliseconds, making it ideal for systems with
low latency requirements. | cs.CL | PATCorrect: Non-autoregressive Phoneme-augmented Transformer for ASR Error Correction |
2023-02-10T03:59:59Z | http://arxiv.org/pdf/2302.05039v1 | http://arxiv.org/abs/2302.05039v1 | Tazin Afrin, Diane Litman | 2023-02-10T03:59:59Z | cs.CL, cs.AI | We develop models to classify desirable evidence and desirable reasoning
revisions in student argumentative writing. We explore two ways to improve
classifier performance - using the essay context of the revision, and using the
feedback students received before the revision. We perform both intrinsic and
extrinsic evaluation for each of our models and report a qualitative analysis.
Our results show that while a model using feedback information improves over a
baseline model, models utilizing context - either alone or with feedback - are
the most successful in identifying desirable revisions. | cs.CL | Predicting Desirable Revisions of Evidence and Reasoning in Argumentative Writing |
2023-02-10T02:55:52Z | http://arxiv.org/pdf/2302.06419v2 | http://arxiv.org/abs/2302.06419v2 | Jiachen Lian, Alexei Baevski, Wei-Ning Hsu, Michael Auli | 2024-01-21T07:41:02Z | eess.AS, cs.AI, cs.CL | Self-supervision has shown great potential for audio-visual speech
recognition by vastly reducing the amount of labeled data required to build
good systems. However, existing methods are either not entirely end-to-end or
do not train joint representations of both modalities. In this paper, we
introduce AV-data2vec which addresses these challenges and builds audio-visual
representations based on predicting contextualized representations which has
been successful in the uni-modal case. The model uses a shared transformer
encoder for both audio and video and can combine both modalities to improve
speech recognition. Results on LRS3 show that AV-data2vec consistently
outperforms existing methods under all settings with the same amount of data
and model size. | eess.AS | AV-data2vec: Self-supervised Learning of Audio-Visual Speech Representations with Contextualized Target Representations |
2023-02-10T02:22:44Z | http://arxiv.org/pdf/2302.05016v2 | http://arxiv.org/abs/2302.05016v2 | Avinash Madasu, Vasudev Lal | 2023-04-15T00:04:54Z | cs.CV, cs.AI, cs.CL | Vision (image and video) - Language (VL) pre-training is the recent popular
paradigm that achieved state-of-the-art results on multi-modal tasks like
image-retrieval, video-retrieval, visual question answering etc. These models
are trained in an unsupervised way and greatly benefit from the complementary
modality supervision. In this paper, we explore if the language representations
trained using vision supervision perform better than vanilla language
representations on Natural Language Understanding and commonsense reasoning
benchmarks. We experiment with a diverse set of image-text models such as
ALBEF, BLIP, METER and video-text models like ALPRO, Frozen-in-Time (FiT),
VIOLET. We compare the performance of language representations of stand-alone
text encoders of these models to the language representations of text encoders
learnt through vision supervision. Our experiments suggest that vanilla
language representations show superior performance on most of the tasks. These
results shed light on the current drawbacks of the vision-language models. | cs.CV | Is Multimodal Vision Supervision Beneficial to Language? |
2023-02-10T01:40:03Z | http://arxiv.org/pdf/2302.06470v1 | http://arxiv.org/abs/2302.06470v1 | Yu Li, Yi Zhang, Weijia Wu, Zimu Zhou, Qiang Li | 2023-02-10T01:40:03Z | cs.CL, Online Insurance Recommendation, Transfer Learning, Data-to-text
Generation | The insurance industry is shifting their sales mode from offline to online,
in expectation to reach massive potential customers in the digitization era.
Due to the complexity and the nature of insurance products, a cost-effective
online sales solution is to exploit chatbot AI to raise customers' attention
and pass those with interests to human agents for further sales. For high
response and conversion rates of customers, it is crucial for the chatbot to
initiate a conversation with personalized opening sentences, which are
generated with user-specific topic selection and ordering. Such personalized
opening sentence generation is challenging because (i) there are limited
historical samples for conversation topic recommendation in online insurance
sales and (ii) existing text generation schemes often fail to support
customized topic ordering based on user preferences. We design POSGen, a
personalized opening sentence generation scheme dedicated for online insurance
sales. It transfers user embeddings learned from auxiliary online user
behaviours to enhance conversation topic recommendation, and exploits a context
management unit to arrange the recommended topics in user-specific ordering for
opening sentence generation. POSGen is deployed on a real-world online
insurance platform. It achieves 2.33x total insurance premium improvement
through a two-month global test. | cs.CL | POSGen: Personalized Opening Sentence Generation for Online Insurance Sales |
2023-02-10T01:34:24Z | http://arxiv.org/pdf/2302.05008v1 | http://arxiv.org/abs/2302.05008v1 | Haoran Xu, Jean Maillard, Vedanuj Goswami | 2023-02-10T01:34:24Z | cs.CL | Multilingual machine translation (MMT) benefits from cross-lingual transfer
but is a challenging multitask optimization problem. This is partly because
there is no clear framework to systematically learn language-specific
parameters. Self-supervised learning (SSL) approaches that leverage large
quantities of monolingual data (where parallel data is unavailable) have shown
promise by improving translation performance as complementary tasks to the MMT
task. However, jointly optimizing SSL and MMT tasks is even more challenging.
In this work, we first investigate how to utilize intra-distillation to learn
more *language-specific* parameters and then show the importance of these
language-specific parameters. Next, we propose a novel but simple SSL task,
concurrent denoising, that co-trains with the MMT task by concurrently
denoising monolingual data on both the encoder and decoder. Finally, we apply
intra-distillation to this co-training approach. Combining these two approaches
significantly improves MMT performance, outperforming three state-of-the-art
SSL methods by a large margin, e.g., 11.3\% and 3.7\% improvement on an
8-language and a 15-language benchmark compared with MASS, respectively | cs.CL | Language-Aware Multilingual Machine Translation with Self-Supervised Learning |
2023-02-10T00:11:19Z | http://arxiv.org/pdf/2302.04985v1 | http://arxiv.org/abs/2302.04985v1 | Xingwei Tan, Gabriele Pergola, Yulan He | 2023-02-10T00:11:19Z | cs.CL, cs.LG | Existing models to extract temporal relations between events lack a
principled method to incorporate external knowledge. In this study, we
introduce Bayesian-Trans, a Bayesian learning-based method that models the
temporal relation representations as latent variables and infers their values
via Bayesian inference and translational functions. Compared to conventional
neural approaches, instead of performing point estimation to find the best set
parameters, the proposed model infers the parameters' posterior distribution
directly, enhancing the model's capability to encode and express uncertainty
about the predictions. Experimental results on the three widely used datasets
show that Bayesian-Trans outperforms existing approaches for event temporal
relation extraction. We additionally present detailed analyses on uncertainty
quantification, comparison of priors, and ablation studies, illustrating the
benefits of the proposed approach. | cs.CL | Event Temporal Relation Extraction with Bayesian Translational Model |
2023-02-09T23:42:30Z | http://arxiv.org/pdf/2302.04981v1 | http://arxiv.org/abs/2302.04981v1 | Salvador Carrión, Francisco Casacuberta | 2023-02-09T23:42:30Z | cs.CL, cs.AI, cs.LG | We present AutoNMT, a framework to streamline the research of seq-to-seq
models by automating the data pipeline (i.e., file management, data
preprocessing, and exploratory analysis), automating experimentation in a
toolkit-agnostic manner, which allows users to use either their own models or
existing seq-to-seq toolkits such as Fairseq or OpenNMT, and finally,
automating the report generation (plots and summaries). Furthermore, this
library comes with its own seq-to-seq toolkit so that users can easily
customize it for non-standard tasks. | cs.CL | AutoNMT: A Framework to Streamline the Research of Seq2Seq Models |
2023-02-09T23:30:49Z | http://arxiv.org/pdf/2302.04975v1 | http://arxiv.org/abs/2302.04975v1 | Nay San, Martijn Bartelds, Blaine Billings, Ella de Falco, Hendi Feriza, Johan Safri, Wawan Sahrozi, Ben Foley, Bradley McDonnell, Dan Jurafsky | 2023-02-09T23:30:49Z | cs.CL | Recent research using pre-trained transformer models suggests that just 10
minutes of transcribed speech may be enough to fine-tune such a model for
automatic speech recognition (ASR) -- at least if we can also leverage vast
amounts of text data (803 million tokens). But is that much text data
necessary? We study the use of different amounts of text data, both for
creating a lexicon that constrains ASR decoding to possible words (e.g. *dogz
vs. dogs), and for training larger language models that bias the system toward
probable word sequences (e.g. too dogs vs. two dogs). We perform experiments
using 10 minutes of transcribed speech from English (for replicating prior
work) and two additional pairs of languages differing in the availability of
supplemental text data: Gronings and Frisian (~7.5M token corpora available),
and Besemah and Nasal (only small lexica available). For all languages, we
found that using only a lexicon did not appreciably improve ASR performance.
For Gronings and Frisian, we found that lexica and language models derived from
'novel-length' 80k token subcorpora reduced the word error rate (WER) to 39% on
average. Our findings suggest that where a text corpus in the upper tens of
thousands of tokens or more is available, fine-tuning a transformer model with
just tens of minutes of transcribed speech holds some promise towards obtaining
human-correctable transcriptions near the 30% WER rule-of-thumb. | cs.CL | Leveraging supplementary text data to kick-start automatic speech recognition system development with limited transcriptions |
2023-02-09T21:15:36Z | http://arxiv.org/pdf/2304.09082v1 | http://arxiv.org/abs/2304.09082v1 | Michael Robinson, Tate Altman, Denley Lam, Letitia W. Li | 2023-02-09T21:15:36Z | cs.PL, cs.CL, cs.IR, 62P30 (Primary), 06A07 (Secondary), D.3.4 | This paper proposes an unsupervised classification method that partitions a
set of files into non-overlapping dialects based upon their behaviors,
determined by messages produced by a collection of programs that consume them.
The pattern of messages can be used as the signature of a particular kind of
behavior, with the understanding that some messages are likely to co-occur,
while others are not. Patterns of messages can be used to classify files into
dialects. A dialect is defined by a subset of messages, called the required
messages. Once files are conditioned upon dialect and its required messages,
the remaining messages are statistically independent.
With this definition of dialect in hand, we present a greedy algorithm that
deduces candidate dialects from a dataset consisting of a matrix of
file-message data, demonstrate its performance on several file formats, and
prove conditions under which it is optimal. We show that an analyst needs to
consider fewer dialects than distinct message patterns, which reduces their
cognitive load when studying a complex format. | cs.PL | Unsupervised clustering of file dialects according to monotonic decompositions of mixtures |
2023-02-09T20:53:12Z | http://arxiv.org/pdf/2302.04931v1 | http://arxiv.org/abs/2302.04931v1 | Mukai Li, Shansan Gong, Jiangtao Feng, Yiheng Xu, Jun Zhang, Zhiyong Wu, Lingpeng Kong | 2023-02-09T20:53:12Z | cs.CL, cs.AI | Large pre-training language models (PLMs) have shown promising in-context
learning abilities. However, due to the backbone transformer architecture,
existing PLMs are bottlenecked by the memory and computational cost when
scaling up to a large context size, leaving instruction tuning and in-context
learning of many demonstration examples, as well as long-range language
modeling under-explored. In this study, we propose a long-range language model
EVALM based on an efficient transformer mechanism. EVALM is trained with 8k
tokens per batch line and can test up to 256k-lengthed contexts with
extrapolation, 128 times to the limit of existing PLMs (e.g. GPT3). Based on
EVALM, we scale up the size of examples efficiently in both instruction tuning
and in-context learning to explore the boundary of the benefits from more
annotated data. Experimental results on a diverse set of tasks show that EVALM
achieves 4.1% higher accuracy on average, and the average length of achieving
the best accuracy score over tasks is around 12k. We find that in-context
learning can achieve higher performance with more demonstrations under
many-shot instruction tuning (8k), and further extending the length of
instructions (16k) can further improve the upper bound of scaling in-context
learning. | cs.CL | In-Context Learning with Many Demonstration Examples |
2023-02-09T19:56:37Z | http://arxiv.org/pdf/2302.04914v3 | http://arxiv.org/abs/2302.04914v3 | Maciej P. Polak, Shrey Modi, Anna Latosinska, Jinming Zhang, Ching-Wen Wang, Shaonan Wang, Ayan Deep Hazra, Dane Morgan | 2024-06-12T14:25:15Z | cond-mat.mtrl-sci, cs.AI, cs.CL | Accurate and comprehensive material databases extracted from research papers
are crucial for materials science and engineering, but their development
requires significant human effort. With large language models (LLMs)
transforming the way humans interact with text, LLMs provide an opportunity to
revolutionize data extraction. In this study, we demonstrate a simple and
efficient method for extracting materials data from full-text research papers
leveraging the capabilities of LLMs combined with human supervision. This
approach is particularly suitable for mid-sized databases and requires minimal
to no coding or prior knowledge about the extracted property. It offers high
recall and nearly perfect precision in the resulting database. The method is
easily adaptable to new and superior language models, ensuring continued
utility. We show this by evaluating and comparing its performance on GPT-3 and
GPT-3.5/4 (which underlie ChatGPT), as well as free alternatives such as BART
and DeBERTaV3. We provide a detailed analysis of the method's performance in
extracting sentences containing bulk modulus data, achieving up to 90%
precision at 96% recall, depending on the amount of human effort involved. We
further demonstrate the method's broader effectiveness by developing a database
of critical cooling rates for metallic glasses over twice the size of previous
human curated databases. | cond-mat.mtrl-sci | Flexible, Model-Agnostic Method for Materials Data Extraction from Text Using General Purpose Language Models |
2023-02-09T19:27:34Z | http://arxiv.org/pdf/2302.04907v1 | http://arxiv.org/abs/2302.04907v1 | Yichi Zhang, Ankush Garg, Yuan Cao, Łukasz Lew, Behrooz Ghorbani, Zhiru Zhang, Orhan Firat | 2023-02-09T19:27:34Z | cs.CL, cs.LG | The rapid scaling of language models is motivating research using
low-bitwidth quantization. In this work, we propose a novel binarization
technique for Transformers applied to machine translation (BMT), the first of
its kind. We identify and address the problem of inflated dot-product variance
when using one-bit weights and activations. Specifically, BMT leverages
additional LayerNorms and residual connections to improve binarization quality.
Experiments on the WMT dataset show that a one-bit weight-only Transformer can
achieve the same quality as a float one, while being 16x smaller in size.
One-bit activations incur varying degrees of quality drop, but mitigated by the
proposed architectural changes. We further conduct a scaling law study using
production-scale translation datasets, which shows that one-bit weight
Transformers scale and generalize well in both in-domain and out-of-domain
settings. Implementation in JAX/Flax will be open sourced. | cs.CL | Binarized Neural Machine Translation |
2023-02-09T18:59:55Z | http://arxiv.org/pdf/2302.04870v1 | http://arxiv.org/abs/2302.04870v1 | Guangxuan Xiao, Ji Lin, Song Han | 2023-02-09T18:59:55Z | cs.CL, cs.CV, cs.LG | Transfer learning is important for foundation models to adapt to downstream
tasks. However, many foundation models are proprietary, so users must share
their data with model owners to fine-tune the models, which is costly and raise
privacy concerns. Moreover, fine-tuning large foundation models is
computation-intensive and impractical for most downstream users. In this paper,
we propose Offsite-Tuning, a privacy-preserving and efficient transfer learning
framework that can adapt billion-parameter foundation models to downstream data
without access to the full model. In offsite-tuning, the model owner sends a
light-weight adapter and a lossy compressed emulator to the data owner, who
then fine-tunes the adapter on the downstream data with the emulator's
assistance. The fine-tuned adapter is then returned to the model owner, who
plugs it into the full model to create an adapted foundation model.
Offsite-tuning preserves both parties' privacy and is computationally more
efficient than the existing fine-tuning methods that require access to the full
model weights. We demonstrate the effectiveness of offsite-tuning on various
large language and vision foundation models. Offsite-tuning can achieve
comparable accuracy as full model fine-tuning while being privacy-preserving
and efficient, achieving 6.5x speedup and 5.6x memory reduction. Code is
available at https://github.com/mit-han-lab/offsite-tuning. | cs.CL | Offsite-Tuning: Transfer Learning without Full Model |
2023-02-09T18:59:41Z | http://arxiv.org/pdf/2302.04865v1 | http://arxiv.org/abs/2302.04865v1 | Ying Shen, Ismini Lourentzou | 2023-02-09T18:59:41Z | cs.CV, cs.AI, cs.CL, cs.LG | The research community has shown increasing interest in designing intelligent
embodied agents that can assist humans in accomplishing tasks. Despite recent
progress on related vision-language benchmarks, most prior work has focused on
building agents that follow instructions rather than endowing agents the
ability to ask questions to actively resolve ambiguities arising naturally in
embodied environments. To empower embodied agents with the ability to interact
with humans, in this work, we propose an Embodied Learning-By-Asking (ELBA)
model that learns when and what questions to ask to dynamically acquire
additional information for completing the task. We evaluate our model on the
TEACH vision-dialog navigation and task completion dataset. Experimental
results show that ELBA achieves improved task performance compared to baseline
models without question-answering capabilities. | cs.CV | Learning by Asking for Embodied Visual Navigation and Task Completion |
2023-02-09T18:59:18Z | http://arxiv.org/pdf/2302.04863v3 | http://arxiv.org/abs/2302.04863v3 | Almog Gueta, Elad Venezian, Colin Raffel, Noam Slonim, Yoav Katz, Leshem Choshen | 2023-10-12T18:42:34Z | cs.LG, cs.AI, cs.CL | Research on neural networks has focused on understanding a single model
trained on a single dataset. However, relatively little is known about the
relationships between different models, particularly those trained or tested on
different datasets. We address this by studying how the weight space and the
underlying loss landscape of different models are interconnected.
Specifically, we demonstrate that finetuned models that were optimized for
high performance, reside in well-defined regions in weight space, and vice
versa -- that any model that resides anywhere in those regions also exhibits
high performance. Notably, we show that language models that have been
finetuned on the same dataset form a tight cluster in the weight space, while
models finetuned on different datasets from the same underlying task form a
looser cluster. Moreover, traversing around the region between the models leads
to new models that perform comparably or even better than models obtained via
finetuning, even on tasks that the original models were not finetuned on.
Our findings provide insight into the relationships between models,
demonstrating that a model positioned between two similar models can acquire
the knowledge of both. We leverage this and design a method for selecting a
better model for efficient finetuning. Specifically, we show that starting from
the center of the region is as effective, if not more, than using the
pretrained model in 11 out of 12 datasets, resulting in an average accuracy
improvement of 3.06. | cs.LG | Knowledge is a Region in Weight Space for Fine-tuned Language Models |
2023-02-09T18:57:56Z | http://arxiv.org/pdf/2302.04858v2 | http://arxiv.org/abs/2302.04858v2 | Zhuolin Yang, Wei Ping, Zihan Liu, Vijay Korthikanti, Weili Nie, De-An Huang, Linxi Fan, Zhiding Yu, Shiyi Lan, Bo Li, Ming-Yu Liu, Yuke Zhu, Mohammad Shoeybi, Bryan Catanzaro, Chaowei Xiao, Anima Anandkumar | 2023-10-22T04:18:00Z | cs.CV, cs.AI, cs.CL, cs.IR, cs.LG | Augmenting pretrained language models (LMs) with a vision encoder (e.g.,
Flamingo) has obtained the state-of-the-art results in image-to-text
generation. However, these models store all the knowledge within their
parameters, thus often requiring enormous model parameters to model the
abundant visual concepts and very rich textual descriptions. Additionally, they
are inefficient in incorporating new data, requiring a computational-expensive
fine-tuning process. In this work, we introduce a Retrieval-augmented Visual
Language Model, Re-ViLM, built upon the Flamingo, that supports retrieving the
relevant knowledge from the external database for zero and in-context few-shot
image-to-text generations. By storing certain knowledge explicitly in the
external database, our approach reduces the number of model parameters and can
easily accommodate new data during evaluation by simply updating the database.
We also construct an interleaved image and text data that facilitates
in-context few-shot learning capabilities. We demonstrate that Re-ViLM
significantly boosts performance for image-to-text generation tasks, especially
for zero-shot and few-shot generation in out-of-domain settings with 4 times
less parameters compared with baseline methods. | cs.CV | Re-ViLM: Retrieval-Augmented Visual Language Model for Zero and Few-Shot Image Captioning |
2023-02-09T18:41:04Z | http://arxiv.org/pdf/2302.04834v1 | http://arxiv.org/abs/2302.04834v1 | Yucheng Li, Shun Wang, Chenghua Lin, Frank Guerin, Loïc Barrault | 2023-02-09T18:41:04Z | cs.CL | In this paper, we propose FrameBERT, a RoBERTa-based model that can
explicitly learn and incorporate FrameNet Embeddings for concept-level metaphor
detection. FrameBERT not only achieves better or comparable performance to the
state-of-the-art, but also is more explainable and interpretable compared to
existing models, attributing to its ability of accounting for external
knowledge of FrameNet. | cs.CL | FrameBERT: Conceptual Metaphor Detection with Frame Embedding Learning |
2023-02-09T18:02:34Z | http://arxiv.org/pdf/2302.04813v3 | http://arxiv.org/abs/2302.04813v3 | Xi Ye, Greg Durrett | 2023-10-18T14:42:15Z | cs.CL | Recent work has shown how to prompt large language models with explanations
to obtain strong performance on textual reasoning tasks, i.e., the
chain-of-thought paradigm. However, subtly different explanations can yield
widely varying downstream task accuracy. Explanations that have not been
"tuned" for a task, such as off-the-shelf explanations written by nonexperts,
may lead to mediocre performance. This paper tackles the problem of how to
optimize explanation-infused prompts in a blackbox fashion. We first generate
sets of candidate explanations for each example in the prompt using a
leave-one-out scheme, then find an effective combination of these explanations
with a two-stage framework. We first evaluate explanations for each in-context
example in isolation according to two proxy metrics, log likelihood and
accuracy on new examples. Then, we search over combinations of explanations to
find one that yields high performance against a silver-labeled development set.
Across four textual reasoning tasks spanning question answering, mathematical
reasoning, and natural language inference, results show that our proxy metrics
correlate with ground truth accuracy and our overall method can effectively
improve prompts over crowdworker annotations and naive search strategies | cs.CL | Explanation Selection Using Unlabeled Data for Chain-of-Thought Prompting |
Subsets and Splits