Search Results for author:
Found 39 papers, 25 papers with code
The False Promise of Imitating Proprietary LLMs
This approach looks to cheaply imitate the proprietary model's capabilities using a weaker open-source model.
AutoPrompt: Eliciting Knowledge from Language Models with Automatically Generated Prompts
The remarkable success of pretrained language models has motivated the study of what kinds of knowledge these models learn during pretraining.
Train Large, Then Compress: Rethinking Model Size for Efficient Training and Inference of Transformers
Since hardware resources are limited, the objective of training deep learning models is typically to maximize accuracy subject to the time and memory constraints of training and inference.
Calibrate Before Use: Improving Few-Shot Performance of Language Models
We show that this type of few-shot learning can be unstable: the choice of prompt format, training examples, and even the order of the training examples can cause accuracy to vary from near chance to near state-of-the-art.
InCoder: A Generative Model for Code Infilling and Synthesis
Our model is the first generative model that is able to directly perform zero-shot code infilling, which we evaluate on challenging tasks such as type inference, comment generation, and variable re-naming.
Ranked #84 on
Code Generation
on MBPP
Universal Adversarial Triggers for Attacking and Analyzing NLP
We define universal adversarial triggers: input-agnostic sequences of tokens that trigger a model to produce a specific prediction when concatenated to any input from a dataset.
Deduplicating Training Data Mitigates Privacy Risks in Language Models
Past work has shown that large language models are susceptible to privacy attacks, where adversaries generate sequences from a trained model and detect which sequences are memorized from the training set.
Extracting Training Data from Large Language Models
We demonstrate our attack on GPT-2, a language model trained on scrapes of the public Internet, and are able to extract hundreds of verbatim text sequences from the model's training data.
Automated Crossword Solving
We present the Berkeley Crossword Solver, a state-of-the-art approach for automatically solving crossword puzzles.
Cutting Down on Prompts and Parameters: Simple Few-Shot Learning with Language Models
Prompting language models (LMs) with training examples and task descriptions has been seen as critical to recent successes in few-shot learning.
SILO Language Models: Isolating Legal Risk In a Nonparametric Datastore
SILO is built by (1) training a parametric LM on Open License Corpus (OLC), a new corpus we curate with 228B tokens of public domain and permissively licensed text and (2) augmenting it with a more general and easily modifiable nonparametric datastore (e. g., containing copyrighted books or news) that is only queried during inference.
Large Language Models Struggle to Learn Long-Tail Knowledge
The Internet contains a wealth of knowledge -- from the birthdays of historical figures to tutorials on how to code -- all of which may be learned by language models.
Pretrained Transformers Improve Out-of-Distribution Robustness
Although pretrained Transformers such as BERT achieve high accuracy on in-distribution examples, do they generalize to new distributions?
Compositional Questions Do Not Necessitate Multi-hop Reasoning
Multi-hop reading comprehension (RC) questions are challenging because they require reading and reasoning over multiple paragraphs.
Evaluating Models' Local Decision Boundaries via Contrast Sets
Unfortunately, when a dataset has systematic gaps (e. g., annotation artifacts), these evaluations are misleading: a model can learn simple decision rules that perform well on the test set but do not capture a dataset's intended capabilities.
Poisoning Language Models During Instruction Tuning
In this work, we show that adversaries can contribute poison examples to these datasets, allowing them to manipulate model predictions whenever a desired trigger phrase appears in the input.
Imitation Attacks and Defenses for Black-box Machine Translation Systems
To mitigate these vulnerabilities, we propose a defense that modifies translation outputs in order to misdirect the optimization of imitation models.
Interpreting Neural Networks With Nearest Neighbors
However, the confidence of neural networks is not a robust measure of model uncertainty.
Trick Me If You Can: Human-in-the-loop Generation of Adversarial Examples for Question Answering
We propose human-in-the-loop adversarial generation, where human authors are guided to break models.
What Evidence Do Language Models Find Convincing?
Retrieval-augmented language models are being increasingly tasked with subjective, contentious, and conflicting queries such as "is aspartame linked to cancer".
Understanding Impacts of High-Order Loss Approximations and Features in Deep Learning Interpretation
Second, we compute the importance of group-features in deep learning interpretation by introducing a sparsity regularization term.
Do NLP Models Know Numbers? Probing Numeracy in Embeddings
The ability to understand and work with numbers (numeracy) is critical for many complex reasoning tasks.
AllenNLP Interpret: A Framework for Explaining Predictions of NLP Models
Neural NLP models are increasingly accurate but are imperfect and opaque---they break in counterintuitive ways and leave end users puzzled at their behavior.
Analyzing Dynamic Adversarial Training Data in the Limit
To create models that are robust across a wide range of test inputs, training datasets should include diverse examples that span numerous phenomena.
Detoxifying Language Models Risks Marginalizing Minority Voices
Language models (LMs) must be both safe and equitable to be responsibly deployed in practice.
Pathologies of Neural Models Make Interpretations Difficult
In existing interpretation methods for NLP, a word's importance is determined by either input perturbation---measuring the decrease in model confidence when that word is removed---or by the gradient with respect to that word.
Trick Me If You Can: Adversarial Writing of Trivia Challenge Questions
Modern question answering systems have been touted as approaching human performance.
Misleading Failures of Partial-input Baselines
Recent work establishes dataset difficulty and removes annotation artifacts via partial-input baselines (e. g., hypothesis-only models for SNLI or question-only models for VQA).
Trustworthy AI Inference Systems: An Industry Research View
Additionally, such systems should also use Privacy-Enhancing Technologies (PETs) to protect customers' data at any time.
Train Big, Then Compress: Rethinking Model Size for Efficient Training and Inference of Transformers
Since hardware resources are limited, the objective of training deep learning models is typically to maximize accuracy subject to the time and memory constraints of training and inference.
Gradient-based Analysis of NLP Models is Manipulable
Gradient-based analysis methods, such as saliency map visualizations and adversarial input perturbations, have found widespread use in interpreting neural NLP models due to their simplicity, flexibility, and most importantly, their faithfulness.
Concealed Data Poisoning Attacks on NLP Models
In this work, we develop a new data poisoning attack that allows an adversary to control model predictions whenever a desired trigger phrase is present in the input.
Evaluating NLP Models via Contrast Sets
Unfortunately, when a dataset has systematic gaps (e. g., annotation artifacts), these evaluations are misleading: a model can learn simple decision rules that perform well on the test set but do not capture a dataset's intended capabilities.
Interpreting Predictions of NLP Models
Although neural NLP models are highly expressive and empirically successful, they also systematically fail in counterintuitive ways and are opaque in their decision-making process.
Measuring Forgetting of Memorized Training Examples
In memorization, models overfit specific training examples and become susceptible to privacy attacks.
Extracting Training Data from Diffusion Models
Image diffusion models such as DALL-E 2, Imagen, and Stable Diffusion have attracted significant attention due to their ability to generate high-quality synthetic images.
Privacy Side Channels in Machine Learning Systems
Most current approaches for protecting privacy in machine learning (ML) assume that models exist in a vacuum, when in reality, ML models are part of larger systems that include components for training data filtering, output monitoring, and more.
Scalable Extraction of Training Data from (Production) Language Models
This paper studies extractable memorization: training data that an adversary can efficiently extract by querying a machine learning model without prior knowledge of the training dataset.
Unfamiliar Finetuning Examples Control How Language Models Hallucinate
Large language models (LLMs) have a tendency to generate plausible-sounding yet factually incorrect responses, especially when queried on unfamiliar concepts.
