Search Results for author:
Found 42 papers, 26 papers with code
Reasoning to Learn from Latent Thoughts
We show that a 1B LM can bootstrap its performance across at least three iterations and significantly outperform baselines trained on raw data, with increasing gains from additional inference compute when performing the E-step.
MixMin: Finding Data Mixtures via Convex Minimization
We formalize this data mixing problem as a bi-level objective: the best mixture is the one that would lead to the best model for a downstream objective.
On the Efficiency of ERM in Feature Learning
Given a collection of feature maps indexed by a set $\mathcal{T}$, we study the performance of empirical risk minimization (ERM) on regression problems with square loss over the union of the linear classes induced by these feature maps.
End-To-End Causal Effect Estimation from Unstructured Natural Language Data
Our results suggest that unstructured text data is a rich source of causal effect information, and NATURAL is a first step towards an automated pipeline to tap this resource.
APPL: A Prompt Programming Language for Harmonious Integration of Programs and Large Language Model Prompts
Large Language Models (LLMs) have become increasingly capable of handling diverse tasks with the aid of well-crafted prompts and integration of external tools, but as task complexity rises, the workflow involving LLMs can be complicated and thus challenging to implement and maintain.
Minimax Linear Regression under the Quantile Risk
We study the problem of designing minimax procedures in linear regression under the quantile risk.
Test-Time Fairness and Robustness in Large Language Models
Finally, we introduce a data augmentation strategy that guarantees stratified invariance at test time under suitable assumptions, together with a prompting strategy that encourages stratified invariance in LLMs.
Finding Optimally Robust Data Mixtures via Concave Maximization
In our experiments, we found that MixMax matched or outperformed the standard group DRO baselines, and in particular, MixMax improved the performance of XGBoost over the only baseline, data balancing, for variations of the ACSIncome and CelebA annotations datasets.
Observational Scaling Laws and the Predictability of Language Model Performance
However, we show that these variations are consistent with a simple, generalized scaling law where language model performance is a function of a low-dimensional capability space, and model families only vary in their efficiency in converting training compute to capabilities.
Experts Don't Cheat: Learning What You Don't Know By Predicting Pairs
Identifying how much a model ${\widehat{p}}_{\theta}(Y|X)$ knows about the stochastic real-world process $p(Y|X)$ it was trained on is important to ensure it avoids producing incorrect or "hallucinated" answers or taking unsafe actions.
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.
Benchmarking Neural Network Training Algorithms
In order to address these challenges, we introduce a new, competitive, time-to-result benchmark using multiple workloads running on fixed hardware, the AlgoPerf: Training Algorithms benchmark.
Contrastive Learning Can Find An Optimal Basis For Approximately View-Invariant Functions
We give generalization bounds for downstream linear prediction using our Kernel PCA representation, and show empirically on a set of synthetic tasks that applying Kernel PCA to contrastive learning models can indeed approximately recover the Markov chain eigenfunctions, although the accuracy depends on the kernel parameterization as well as on the augmentation strength.
Learning To Cut By Looking Ahead: Cutting Plane Selection via Imitation Learning
Cutting planes are essential for solving mixed-integer linear problems (MILPs), because they facilitate bound improvements on the optimal solution value.
The Machine Learning for Combinatorial Optimization Competition (ML4CO): Results and Insights
Combinatorial optimization is a well-established area in operations research and computer science.
Augment with Care: Contrastive Learning for Combinatorial Problems
While typical graph contrastive pre-training uses label-agnostic augmentations, our key insight is that many combinatorial problems have well-studied invariances, which allow for the design of label-preserving augmentations.
Bayesian Nonparametrics for Offline Skill Discovery
Skills or low-level policies in reinforcement learning are temporally extended actions that can speed up learning and enable complex behaviours.
Optimal Representations for Covariate Shift
Machine learning systems often experience a distribution shift between training and testing.
Ranked #38 on
Image Classification
on ObjectNet
(using extra training data)
Learning Generalized Gumbel-max Causal Mechanisms
To perform counterfactual reasoning in Structural Causal Models (SCMs), one needs to know the causal mechanisms, which provide factorizations of conditional distributions into noise sources and deterministic functions mapping realizations of noise to samples.
Unbiased Gradient Estimation with Balanced Assignments for Mixtures of Experts
Training large-scale mixture of experts models efficiently on modern hardware requires assigning datapoints in a batch to different experts, each with a limited capacity.
Lossy Compression for Lossless Prediction
Most data is automatically collected and only ever "seen" by algorithms.
Ranked #1 on
Image Compression
on Oxford-IIIT Pet Dataset
(using extra training data)
Learning to Extend Program Graphs to Work-in-Progress Code
Source code spends most of its time in a broken or incomplete state during software development.
Improving Lossless Compression Rates via Monte Carlo Bits-Back Coding
Naively applied, our schemes would require more initial bits than the standard bits-back coder, but we show how to drastically reduce this additional cost with couplings in the latent space.
Oops I Took A Gradient: Scalable Sampling for Discrete Distributions
We propose a general and scalable approximate sampling strategy for probabilistic models with discrete variables.
Rao-Blackwellizing the Straight-Through Gumbel-Softmax Gradient Estimator
Gradient estimation in models with discrete latent variables is a challenging problem, because the simplest unbiased estimators tend to have high variance.
Learning Branching Heuristics for Propositional Model Counting
In addition to step count improvements, Neuro# can also achieve orders of magnitude wall-clock speedups over the vanilla solver on larger instances in some problem families, despite the runtime overhead of querying the model.
Gradient Estimation with Stochastic Softmax Tricks
The Gumbel-Max trick is the basis of many relaxed gradient estimators.
On Empirical Comparisons of Optimizers for Deep Learning
In particular, we find that the popular adaptive gradient methods never underperform momentum or gradient descent.
Direct Policy Gradients: Direct Optimization of Policies in Discrete Action Spaces
A main benefit of DirPG algorithms is that they allow the insertion of domain knowledge in the form of upper bounds on return-to-go at training time, like is used in heuristic search, while still directly computing a policy gradient.
Continuous Hierarchical Representations with Poincaré Variational Auto-Encoders
We therefore endow VAEs with a Poincar\'e ball model of hyperbolic geometry as a latent space and rigorously derive the necessary methods to work with two main Gaussian generalisations on that space.
Doubly Reparameterized Gradient Estimators for Monte Carlo Objectives
Burda et al. (2015) introduced a multi-sample variational bound, IWAE, that is at least as tight as the standard variational lower bound and becomes increasingly tight as the number of samples increases.
Hamiltonian Descent Methods
Yet, crucially the kinetic gradient map can be designed to incorporate information about the convex conjugate in a fashion that allows for linear convergence on convex functions that may be non-smooth or non-strongly convex.
Conditional Neural Processes
Deep neural networks excel at function approximation, yet they are typically trained from scratch for each new function.
Tighter Variational Bounds are Not Necessarily Better
We provide theoretical and empirical evidence that using tighter evidence lower bounds (ELBOs) can be detrimental to the process of learning an inference network by reducing the signal-to-noise ratio of the gradient estimator.
Filtering Variational Objectives
When used as a surrogate objective for maximum likelihood estimation in latent variable models, the evidence lower bound (ELBO) produces state-of-the-art results.
REBAR: Low-variance, unbiased gradient estimates for discrete latent variable models
Learning in models with discrete latent variables is challenging due to high variance gradient estimators.
Particle Value Functions
The policy gradients of the expected return objective can react slowly to rare rewards.
The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables
The essence of the trick is to refactor each stochastic node into a differentiable function of its parameters and a random variable with fixed distribution.
Move Evaluation in Go Using Deep Convolutional Neural Networks
The game of Go is more challenging than other board games, due to the difficulty of constructing a position or move evaluation function.
A* Sampling
The problem of drawing samples from a discrete distribution can be converted into a discrete optimization problem.
Structured Generative Models of Natural Source Code
We study the problem of building generative models of natural source code (NSC); that is, source code written and understood by humans.
Annealing between distributions by averaging moments
Many powerful Monte Carlo techniques for estimating partition functions, such as annealed importance sampling (AIS), are based on sampling from a sequence of intermediate distributions which interpolate between a tractable initial distribution and an intractable target distribution.
