1 |
Probing Classifiers: Promises, Shortcomings, and Advances ...
|
|
|
|
BASE
|
|
Show details
|
|
2 |
On the Pitfalls of Analyzing Individual Neurons in Language Models ...
|
|
|
|
BASE
|
|
Show details
|
|
3 |
Debiasing Methods in Natural Language Understanding Make Bias More Accessible ...
|
|
|
|
BASE
|
|
Show details
|
|
4 |
Causal Analysis of Syntactic Agreement Mechanisms in Neural Language Models ...
|
|
|
|
BASE
|
|
Show details
|
|
5 |
Similarity Analysis of Contextual Word Representation Models ...
|
|
|
|
BASE
|
|
Show details
|
|
6 |
Probing the Probing Paradigm: Does Probing Accuracy Entail Task Relevance? ...
|
|
|
|
BASE
|
|
Show details
|
|
7 |
The Sensitivity of Language Models and Humans to Winograd Schema Perturbations ...
|
|
|
|
BASE
|
|
Show details
|
|
8 |
Analyzing Individual Neurons in Pre-trained Language Models ...
|
|
|
|
BASE
|
|
Show details
|
|
9 |
On the Linguistic Representational Power of Neural Machine Translation Models
|
|
|
|
In: Computational Linguistics, Vol 46, Iss 1, Pp 1-52 (2020) (2020)
|
|
BASE
|
|
Show details
|
|
11 |
Exploring Compositional Architectures and Word Vector Representations for Prepositional Phrase Attachment
|
|
|
|
In: MIT Press (2019)
|
|
BASE
|
|
Show details
|
|
12 |
On the Linguistic Representational Power of Neural Machine Translation Models ...
|
|
|
|
BASE
|
|
Show details
|
|
13 |
On Adversarial Removal of Hypothesis-only Bias in Natural Language Inference ...
|
|
|
|
BASE
|
|
Show details
|
|
14 |
Improving Neural Language Models by Segmenting, Attending, and Predicting the Future ...
|
|
|
|
BASE
|
|
Show details
|
|
15 |
On Adversarial Removal of Hypothesis-only Bias in Natural Language Inference
|
|
|
|
BASE
|
|
Show details
|
|
16 |
On Evaluating the Generalization of LSTM Models in Formal Languages
|
|
|
|
BASE
|
|
Show details
|
|
17 |
Don't Take the Premise for Granted: Mitigating Artifacts in Natural Language Inference
|
|
|
|
Abstract:
Natural Language Inference (NLI) datasets often contain hypothesis-only biases—artifacts that allow models to achieve non-trivial performance without learning whether a premise entails a hypothesis. We propose two probabilistic methods to build models that are more robust to such biases and better transfer across datasets. In contrast to standard approaches to NLI, our methods predict the probability of a premise given a hypothesis and NLI label, discouraging models from ignoring the premise. We evaluate our methods on synthetic and existing NLI datasets by training on datasets containing biases and testing on datasets containing no (or different) hypothesis-only biases. Our results indicate that these methods can make NLI models more robust to dataset-specific artifacts, transferring better than a baseline architecture in 9 out of 12 NLI datasets. Additionally, we provide an extensive analysis of the interplay of our methods with known biases in NLI datasets, as well as the effects of encouraging models to ignore biases and fine-tuning on target datasets. ; Accepted Manuscript
|
|
URL: http://nrs.harvard.edu/urn-3:HUL.InstRepos:40827357
|
|
BASE
|
|
Hide details
|
|
19 |
On Evaluating the Generalization of LSTM Models in Formal Languages
|
|
|
|
In: Proceedings of the Society for Computation in Linguistics (2019)
|
|
BASE
|
|
Show details
|
|
20 |
Analysis Methods in Neural Language Processing: A Survey
|
|
|
|
In: Transactions of the Association for Computational Linguistics, Vol 7, Pp 49-72 (2019) (2019)
|
|
BASE
|
|
Show details
|
|
|
|