2 |
A Neighbourhood Framework for Resource-Lean Content Flagging ...
|
|
|
|
BASE
|
|
Show details
|
|
3 |
A Primer on Contrastive Pretraining in Language Processing: Methods, Lessons Learned and Perspectives ...
|
|
|
|
BASE
|
|
Show details
|
|
4 |
QA Dataset Explosion: A Taxonomy of NLP Resources for Question Answering and Reading Comprehension ...
|
|
|
|
BASE
|
|
Show details
|
|
5 |
Quantifying Gender Bias Towards Politicians in Cross-Lingual Language Models ...
|
|
|
|
BASE
|
|
Show details
|
|
6 |
Few-Shot Cross-Lingual Stance Detection with Sentiment-Based Pre-Training ...
|
|
|
|
BASE
|
|
Show details
|
|
7 |
Can Edge Probing Tasks Reveal Linguistic Knowledge in QA Models? ...
|
|
|
|
Abstract:
There have been many efforts to try to understand what gram-matical knowledge (e.g., ability to understand the part of speech of a token) is encoded in large pre-trained language models (LM). This is done through 'Edge Probing' (EP) tests: simple ML models that predict the grammatical properties ofa span (whether it has a particular part of speech) using only the LM's token representations. However, most NLP applications use fine-tuned LMs. Here, we ask: if a LM is fine-tuned, does the encoding of linguistic information in it change, as measured by EP tests? Conducting experiments on multiple question-answering (QA) datasets, we answer that question negatively: the EP test results do not change significantly when the fine-tuned QA model performs well or in adversarial situations where the model is forced to learn wrong correlations. However, a critical analysis of the EP task datasets reveals that EP models may rely on spurious correlations to make predictions. This indicates even if fine-tuning changes the ...
|
|
Keyword:
Computation and Language cs.CL; FOS Computer and information sciences
|
|
URL: https://arxiv.org/abs/2109.07102 https://dx.doi.org/10.48550/arxiv.2109.07102
|
|
BASE
|
|
Hide details
|
|
8 |
CiteWorth: Cite-Worthiness Detection for Improved Scientific Document Understanding ...
|
|
|
|
BASE
|
|
Show details
|
|
9 |
How Does Counterfactually Augmented Data Impact Models for Social Computing Constructs? ...
|
|
|
|
BASE
|
|
Show details
|
|
12 |
Quantifying Gender Biases Towards Politicians on Reddit ...
|
|
|
|
BASE
|
|
Show details
|
|
14 |
Semi-Supervised Exaggeration Detection of Health Science Press Releases ...
|
|
|
|
BASE
|
|
Show details
|
|
15 |
Inducing Language-Agnostic Multilingual Representations ...
|
|
|
|
BASE
|
|
Show details
|
|
17 |
SIGTYP 2020 Shared Task: Prediction of Typological Features ...
|
|
|
|
BASE
|
|
Show details
|
|
19 |
X-WikiRE: A Large, Multilingual Resource for Relation Extraction as Machine Comprehension ...
|
|
|
|
BASE
|
|
Show details
|
|
20 |
TX-Ray: Quantifying and Explaining Model-Knowledge Transfer in (Un-)Supervised NLP ...
|
|
|
|
BASE
|
|
Show details
|
|
|
|