Search Results for author:
Found 45 papers, 27 papers with code
Removing RLHF Protections in GPT-4 via Fine-Tuning
In tandem, LLM vendors have been increasingly enabling fine-tuning of their most powerful models.
MoCa: Measuring Human-Language Model Alignment on Causal and Moral Judgment Tasks
A rich literature in cognitive science has studied people's causal and moral intuitions.
On the Fairness ROAD: Robust Optimization for Adversarial Debiasing
In the field of algorithmic fairness, significant attention has been put on group fairness criteria, such as Demographic Parity and Equalized Odds.
Learning to (Learn at Test Time)
Our inner loop turns out to be equivalent to linear attention when the inner-loop learner is only a linear model, and to self-attention when it is a kernel estimator.
Benchmarking and Improving Generator-Validator Consistency of Language Models
To improve the consistency of LMs, we propose to finetune on the filtered generator and validator responses that are GV-consistent, and call this approach consistency fine-tuning.
Identifying the Risks of LM Agents with an LM-Emulated Sandbox
Alongside the emulator, we develop an LM-based automatic safety evaluator that examines agent failures and quantifies associated risks.
Safety-Tuned LLaMAs: Lessons From Improving the Safety of Large Language Models that Follow Instructions
Training large language models to follow instructions makes them perform better on a wide range of tasks, generally becoming more helpful.
Identifying and Mitigating the Security Risks of Generative AI
However, GenAI can be used just as well by attackers to generate new attacks and increase the velocity and efficacy of existing attacks.
Where's the Liability in Harmful AI Speech?
We argue that AI should not be categorically immune from liability in these scenarios and that as courts grapple with the already fine-grained complexities of platform algorithms, the technical details of generative AI loom above with thornier questions.
Robust Distortion-free Watermarks for Language Models
We generate watermarked text by mapping a sequence of random numbers -- which we compute using a randomized watermark key -- to a sample from the language model.
Whose Opinions Do Language Models Reflect?
Language models (LMs) are increasingly being used in open-ended contexts, where the opinions reflected by LMs in response to subjective queries can have a profound impact, both on user satisfaction, as well as shaping the views of society at large.
Navigating the Grey Area: How Expressions of Uncertainty and Overconfidence Affect Language Models
The increased deployment of LMs for real-world tasks involving knowledge and facts makes it important to understand model epistemology: what LMs think they know, and how their attitudes toward that knowledge are affected by language use in their inputs.
Out-of-Domain Robustness via Targeted Augmentations
Models trained on one set of domains often suffer performance drops on unseen domains, e. g., when wildlife monitoring models are deployed in new camera locations.
Exploiting Programmatic Behavior of LLMs: Dual-Use Through Standard Security Attacks
Recent advances in instruction-following large language models (LLMs) have led to dramatic improvements in a range of NLP tasks.
Evaluating Self-Supervised Learning via Risk Decomposition
Our decomposition consists of four error components: approximation, representation usability, probe generalization, and encoder generalization.
Contrastive Error Attribution for Finetuned Language Models
We show that existing approaches for error tracing, such as gradient-based influence measures, do not perform reliably for detecting faithfulness errors in NLG datasets.
Privacy-Preserving Domain Adaptation of Semantic Parsers
Task-oriented dialogue systems often assist users with personal or confidential matters.
Holistic Evaluation of Language Models
We present Holistic Evaluation of Language Models (HELM) to improve the transparency of language models.
Easily Accessible Text-to-Image Generation Amplifies Demographic Stereotypes at Large Scale
For example, we find cases of prompting for basic traits or social roles resulting in images reinforcing whiteness as ideal, prompting for occupations resulting in amplification of racial and gender disparities, and prompting for objects resulting in reification of American norms.
Contrastive Decoding: Open-ended Text Generation as Optimization
We propose contrastive decoding (CD), a reliable decoding approach that optimizes a contrastive objective subject to a plausibility constraint.
A Closer Look at the Calibration of Differentially Private Learners
Across 7 tasks, temperature scaling and Platt scaling with DP-SGD result in an average 3. 1-fold reduction in the in-domain expected calibration error and only incur at most a minor percent drop in accuracy.
Improving Self-Supervised Learning by Characterizing Idealized Representations
For non-contrastive learning, we use our framework to derive a simple and novel objective.
Is a Caption Worth a Thousand Images? A Controlled Study for Representation Learning
The development of CLIP [Radford et al., 2021] has sparked a debate on whether language supervision can result in vision models with more transferable representations than traditional image-only methods.
When Does Differentially Private Learning Not Suffer in High Dimensions?
Large pretrained models can be privately fine-tuned to achieve performance approaching that of non-private models.
Emergent Abilities of Large Language Models
Scaling up language models has been shown to predictably improve performance and sample efficiency on a wide range of downstream tasks.
Undersampling is a Minimax Optimal Robustness Intervention in Nonparametric Classification
While a broad range of techniques have been proposed to tackle distribution shift, the simple baseline of training on an $\textit{undersampled}$ balanced dataset often achieves close to state-of-the-art-accuracy across several popular benchmarks.
Spurious Correlations in Reference-Free Evaluation of Text Generation
Model-based, reference-free evaluation metrics have been proposed as a fast and cost-effective approach to evaluate Natural Language Generation (NLG) systems.
Distributionally Robust Models with Parametric Likelihood Ratios
As machine learning models are deployed ever more broadly, it becomes increasingly important that they are not only able to perform well on their training distribution, but also yield accurate predictions when confronted with distribution shift.
Language modeling via stochastic processes
Recent work in self-supervised learning suggests that models can learn good latent representations via contrastive learning, which can be effective for discriminative tasks.
Jury Learning: Integrating Dissenting Voices into Machine Learning Models
We introduce jury learning, a supervised ML approach that resolves these disagreements explicitly through the metaphor of a jury: defining which people or groups, in what proportion, determine the classifier's prediction.
Is Importance Weighting Incompatible with Interpolating Classifiers?
As a remedy, we show that polynomially-tailed losses restore the effects of importance reweighting in correcting distribution shift in overparameterized models.
Extending the WILDS Benchmark for Unsupervised Adaptation
Unlabeled data can be a powerful point of leverage for mitigating these distribution shifts, as it is frequently much more available than labeled data and can often be obtained from distributions beyond the source distribution as well.
Large Language Models Can Be Strong Differentially Private Learners
Differentially Private (DP) learning has seen limited success for building large deep learning models of text, and straightforward attempts at applying Differentially Private Stochastic Gradient Descent (DP-SGD) to NLP tasks have resulted in large performance drops and high computational overhead.
On the Opportunities and Risks of Foundation Models
AI is undergoing a paradigm shift with the rise of models (e. g., BERT, DALL-E, GPT-3) that are trained on broad data at scale and are adaptable to a wide range of downstream tasks.
Proof: Accelerating Approximate Aggregation Queries with Expensive Predicates
Given a dataset $\mathcal{D}$, we are interested in computing the mean of a subset of $\mathcal{D}$ which matches a predicate.
Measuring Conversational Uptake: A Case Study on Student-Teacher Interactions
In conversation, uptake happens when a speaker builds on the contribution of their interlocutor by, for example, acknowledging, repeating or reformulating what they have said.
On the Inductive Bias of Masked Language Modeling: From Statistical to Syntactic Dependencies
We study how masking and predicting tokens in an unsupervised fashion can give rise to linguistic structures and downstream performance gains.
Modeling the Second Player in Distributionally Robust Optimization
Distributionally robust optimization (DRO) provides a framework for training machine learning models that are able to perform well on a collection of related data distributions (the "uncertainty set").
The GEM Benchmark: Natural Language Generation, its Evaluation and Metrics
We introduce GEM, a living benchmark for natural language Generation (NLG), its Evaluation, and Metrics.
Ranked #1 on
Extreme Summarization
on GEM-XSum
Abstractive Text Summarization
Cross-Lingual Abstractive Summarization
+5
Predicting the impact of dataset composition on model performance
Real-world machine learning systems are often are trained using a mix of data sources with varying cost and quality.
Distributionally Robust Losses for Latent Covariate Mixtures
While modern large-scale datasets often consist of heterogeneous subpopulations -- for example, multiple demographic groups or multiple text corpora -- the standard practice of minimizing average loss fails to guarantee uniformly low losses across all subpopulations.
Robustness to Spurious Correlations via Human Annotations
The reliability of machine learning systems critically assumes that the associations between features and labels remain similar between training and test distributions.
Improved Natural Language Generation via Loss Truncation
In this work, we show that the distinguishability of the models and reference serves as a principled and robust alternative for handling invalid references.
Inferring Multidimensional Rates of Aging from Cross-Sectional Data
Motivated by the study of human aging, we present an interpretable latent-variable model that learns temporal dynamics from cross-sectional data.
