Text-to-image models suffer from various safety issues that may limit th...
Much of the knowledge encoded in transformer language models (LMs) may b...
Recent studies show that instruction tuning and learning from human feed...
Before deploying a language model (LM) within a given domain, it is impo...
Text-to-image models are trained on extensive amounts of data, leading t...
Mathematical reasoning in large language models (LLMs) has garnered atte...
Recent advances in interpretability suggest we can project weights and h...
Natural language processing models tend to learn and encode social biase...
Recent work has compared neural network representations via similarity-b...
Text-to-image diffusion models often make implicit assumptions about the...
For applications that require processing large amounts of text at infere...
Models trained from real-world data tend to imitate and amplify social
b...
Dual encoders are now the dominant architecture for dense retrieval. Yet...
The field of emergent communication aims to understand the characteristi...
Considerable efforts to measure and mitigate gender bias in recent years...
Neural networks are known to exploit spurious artifacts (or shortcuts) t...
Recent work has shown exciting promise in updating large language models...
Large amounts of training data are one of the major reasons for the high...
Large pre-trained models are usually fine-tuned on downstream task data,...
Huge language models (LMs) have ushered in a new era for AI, serving as ...
Common studies of gender bias in NLP focus either on extrinsic bias meas...
Most evaluations of attribution methods focus on the English language. I...
We investigate the mechanisms underlying factual knowledge recall in
aut...
While many studies have shown that linguistic information is encoded in
...
Model robustness to bias is often determined by the generalization on
ca...
Many natural language inference (NLI) datasets contain biases that allow...
Targeted syntactic evaluations have demonstrated the ability of language...
While large-scale pretrained language models have obtained impressive re...
Natural Language Inference (NLI) models are known to learn from biases a...
Probing classifiers have emerged as one of the prominent methodologies f...
State-of-the-art natural language processing (NLP) models often learn to...
Self-supervised speech representation learning has recently been a prosp...
While a lot of analysis has been carried to demonstrate linguistic knowl...
The predominant approach to open-domain dialog generation relies on
end-...
Large-scale pretrained language models are the major driving force behin...
This paper investigates contextual word representation models from the l...
Common methods for interpreting neural models in natural language proces...
Large pre-trained contextual word representations have transformed the f...
We introduce three memory-augmented Recurrent Neural Networks (MARNNs) a...
Despite the recent success of deep neural networks in natural language
p...
The dependency of the generalization error of neural networks on model a...
Popular Natural Language Inference (NLI) datasets have been shown to be
...
Natural Language Inference (NLI) datasets often contain hypothesis-only
...
End-to-end neural network systems for automatic speech recognition (ASR)...
We share the findings of the first shared task on improving robustness o...
Visual question answering (VQA) models have been shown to over-rely on
l...
In this paper, we systematically assess the ability of standard recurren...
The Transformer is a fully attention-based alternative to recurrent netw...
Common language models typically predict the next word given the context...
Contextual word representations derived from large-scale neural language...