New phases identified in neural scaling laws with compute limits.
problem Understanding neural scaling laws under compute constraints.
method Solved neural scaling model with stochastic gradient descent, derived loss curves, analyzed model-parameter-count phases.
result Identified 4 phases (+3 subphases) in data-complexity/target-complexity phase-plane, derived exponents.
Estimates model performance from compute budget for distillation.
problem Risk mitigation in large-scale distillation.
method Distillation scaling law based on compute budget allocation.
result Maximizes student performance with compute-optimal allocation.
Memory-efficient learning for large-scale imaging systems.
problem Memory limitations in GPUs for real-world large-scale inverse problems.
method Exploits reversibility of network layers to enable data-driven design.
result Demonstrated on small-scale and large-scale real-world systems.
PALMS reconstructs large-scale networks efficiently with parallel computing.
problem Reconstructing large-scale latent networks from observed dynamics is computationally challenging.
method PALMS (Parallel Adaptive Lasso with Multi-directional Signals) framework for distributed network reconstruction.
result PALMS substantially reduces computational complexity and storage requirements.
Scaling laws found for reinforcement learning performance with model size and compute.
problem Challenges in extending generative modeling scaling laws to reinforcement learning.
method Introduced intrinsic performance as a monotonic function of mean episode return.
result Intrinsic performance scales as a power law in model size and environment interactions.
Sloth predicts LLM performance using latent skills across families.
problem Variations in benchmark performance due to differences in training configurations and data processing across model families.
method Sloth uses publicly available benchmark data and assumes LLM performance is driven by latent skills influenced by model size and training tokens. It exploits correlations across benchmarks to provide accurate predictions.
result Sloth predicts LLM performance accurately and offers insights into scaling behaviors for complex tasks.
Deep learning (DL) creates impactful advances following a virtuous recipe: model architecture search, creating large training data sets, and scaling computation. It is widely believed that growing training sets and models should improve accuracy and result in better products. As DL application domains grow, we would li…
The paper predicts loss scaling across different datasets and compute scales.
problem Predicting loss scaling across different datasets and compute scales.
method Derive shifted power law relationships between train and test losses.
result Shifted power law relationships hold for various datasets and tasks, improving prediction accuracy.
The study examines how verifier imperfections impact test-time scaling techniques.
problem Understanding how verifier imperfections affect test-time scaling methods.
method Proves the instance-level accuracy of Best-of-N and Rejection Sampling methods using the geometry of the verifier's ROC curve.
result RS outperforms BoN for fixed compute, but both converge to the same accuracy in the infinite-compute limit.
Novel algorithm speeds up log-determinant estimation for large matrices.
problem Efficiently estimating log-determinants of large positive definite matrices under memory constraints.
method Hierarchical algorithm based on block-wise computation of LDL decomposition.
result Accurate estimation of NTK log-determinants from a tiny fraction of the full dataset.
Simple policy outperforms complex ones in cloud auto-scaling.
problem Predicting resource scaling for large-scale cloud applications with limited deployment throughput.
method Probabilistic workload forecast for auto-scaling decisions based on risk aversion.
result The proposed policy outperforms sophisticated and simple benchmark policies in real-world and synthetic data.
Efficiently solves large-scale robust portfolio optimization problems.
problem High computational demands in large-scale robust portfolio optimization.
method Extended supporting hyperplane approximation for distributionally robust portfolio problems.
result Significantly reduces computational time from several thousand seconds to just a few.
GShard enables scaling of large neural networks with automatic sharding and lightweight APIs.
problem Scaling neural networks to handle vast training data and compute efficiently.
method GShard uses lightweight annotation APIs and XLA compiler extensions for parallel computation.
result GShard successfully trained a 600 billion parameter model on 2048 TPUs in 4 days.
Model predicts neural network performance scaling laws across various factors.
problem Understanding the performance of neural networks across different training factors.
method Random feature model trained with gradient descent, analyzing compute-optimal scaling laws.
result Predicts asymmetric compute-optimal scaling rule and behavior of training and test loss gap.
Quantum computing offers energy savings over classical computing.
problem Energy efficiency in computing services.
method Cournot competition model constrained by energy usage.
result Quantum computing firms can outperform classical counterparts in energy efficiency.
Investigates ways to train larger models with fewer resources, finding that test loss depends only on the actual number of trainable parameters.
problem Training larger models for cheaper under hardware constraints.
method Emulates an increase in effective parameters using frozen random parameters or fast structured transforms.
result Scaling laws cannot be deceived by spurious parameters; test loss depends only on the actual number of trainable parameters.
Quantum machine learning tackles large datasets with randomized measurements.
problem Efficiently process large, high-dimensional datasets on quantum computers.
method Randomized measurements to scale linearly with dataset size and quadratic for post-processing.
result Substantial speed-up for noisy quantum computers, enabling image classification.
Develops active learning for scale-bridging simulations.
problem Quantitative predictions in nanoporous media and inertial confinement fusion.
method Active learning approach to optimize fine-scale simulations for coarse-scale hydrodynamics.
result Optimizes use of fine-scale simulations for coarse-scale predictions.
This work improves Gaussian process model selection for large datasets.
problem Prohibitively high computational cost in Gaussian process model selection.
method Linear-time scaling and computational uncertainty tradeoff.
result Computation-aware Gaussian processes can be trained on large datasets efficiently.
Predicts language model performance from public models without training.
problem Understanding how language model performance varies with scale.
method Observational approach building scaling laws from publicly available models.
result Smooth, sigmoidal behavior of emergent phenomena is predictable from small models.
Efficiently scales continuous kernels with sparse Fourier domain learning.
problem High computational and memory demands, spectral bias in continuous kernels.
method Sparse learning in the Fourier domain.
result Efficient scaling of continuous kernels, reduced computational and memory requirements, mitigated spectral bias.
New framework explains fast transfer of hyperparameters across model scales.
problem Understanding and optimizing hyperparameters for large-scale models.
method Developed a conceptual framework for HP transfer across scale, showing fast transfer is equivalent to useful transfer for compute-optimal grid search.
result Fast transfer of hyperparameters is equivalent to useful transfer for compute-optimal grid search, offering asymptotic computational advantage.
Scale of data and scale of computation infrastructures together enable the current deep learning renaissance. However, training large-scale deep architectures demands both algorithmic improvement and careful system configuration. In this paper, we focus on employing the system approach to speed up large-scale training.…
New method speeds up analysis of computer experiments.
problem Computational infeasibility of direct GP inference for large datasets.
method Adapted Vecchia's ordered conditional approximation to scaled input space.
result Significant performance improvement over existing methods.
New method improves Kalman filtering and smoothing for large state spaces.
problem High computational cost and uncertainty in large-scale Kalman filtering.
method Probabilistic numerical method leveraging GPU acceleration and tunable trade-off.
result Mitigates scaling issues and provides more accurate uncertainty estimates.
This paper surveys large-scale machine learning methods for efficient data analysis.
problem Efficiently processing large-scale data with machine learning models.
method Divided into three categories: model simplification, optimization approximation, and computation parallelism.
result Blueprint for future developments in large-scale machine learning.
We propose a practical and scalable Gaussian process model for large-scale nonlinear probabilistic regression. Our mixture-of-experts model is conceptually simple and hierarchically recombines computations for an overall approximation of a full Gaussian process. Closed-form and distributed computations allow for effici…
Efficiently selects nearest neighbors for labeling to speed up active learning.
problem Intractable active learning and search for large-scale unlabeled data.
method Restricts candidate pool to nearest neighbors of labeled set.
result Achieved similar performance to global approach but reduced computational cost by up to 3 orders of magnitude.
New algorithm improves inference-time alignment without reward hacking.
problem Improving quality of responses from language models with limited compute.
method Inference-time alignment, focusing on extttInferenceTimePessimism algorithm. result Optimal performance and scaling-monotonicity of extttInferenceTimePessimism. New method predicts and optimizes test-time scaling for LLMs.
problem Lack of principled guidance on scaling LLMs efficiently.
method Tail-guided search to predict and allocate compute.
result SLG Search achieves higher rewards with less compute.
We propose the use of incomplete dot products (IDP) to dynamically adjust the number of input channels used in each layer of a convolutional neural network during feedforward inference. IDP adds monotonically non-increasing coefficients, referred to as a "profile", to the channels during training. The profile orders th…
Model shows feature learning can improve neural scaling laws for hard tasks.
problem Understanding and improving neural network scaling laws for various task difficulties.
method Developed a solvable model of neural scaling laws, identified three scaling regimes, and demonstrated feature learning's impact on scaling exponents.
result Feature learning can improve scaling with training time and compute for hard tasks, nearly doubling the exponent.
Chinchilla Approach 2 biases neural scaling law estimates, leading to unnecessary compute costs.
problem Systematic biases in Chinchilla Approach 2's parabolic fits of neural scaling laws.
method Analyzes three sources of error: IsoFLOP sampling grid width, uncentered sampling, and loss surface asymmetry.
result Chinchilla Approach 3 largely eliminates these biases, offering a more convenient or scalable alternative.
This study explores training Bayesian neural networks at scale using high-performance computing.
problem Challenges in training Bayesian neural networks at scale due to computational overhead.
method High-performance computing with distributed training, network pruning.
result Pruning up to 80% of the network can reduce inference time by 7.0% without significant accuracy loss.
We study empirical scaling laws for language model performance on the cross-entropy loss. The loss scales as a power-law with model size, dataset size, and the amount of compute used for training, with some trends spanning more than seven orders of magnitude. Other architectural details such as network width or depth h…
Encoding the scale information explicitly into the representation learned by a convolutional neural network (CNN) is beneficial for many computer vision tasks especially when dealing with multiscale inputs. We study, in this paper, a scaling-translation-equivariant (ST-equivariant) CNN with joint convolutions across th…
Improved neural framework for scaling entropic MOT with significant computational gains.
problem High computational overhead in multimarginal optimal transport.
method Neural Entropic MOT (NEMOT) using mini-batch training to reduce complexity.
result Significant speedups and feasibility improvements for multimarginal data.
Model shows loss curve with two distinct exponents due to sparse activations.
problem Sparse activations impact neural network scaling laws.
method Introduced a model for neural scaling laws under sparse activations, derived asymptotic population loss, and analyzed gradient-descent dynamics.
result Loss curve exhibits double-descent peak near interpolation threshold with two distinct scaling exponents.
To scale Gaussian processes (GPs) to large data sets we introduce the robust Bayesian Committee Machine (rBCM), a practical and scalable product-of-experts model for large-scale distributed GP regression. Unlike state-of-the-art sparse GP approximations, the rBCM is conceptually simple and does not rely on inducing or …
The theory of quantum computation can be constructed from the abstract study of anyonic systems. In mathematical terms, these are unitary topological modular functors. They underlie the Jones polynomial and arise in Witten-Chern-Simons theory. The braiding and fusion of anyonic excitations in quantum Hall electron liqu…
Representations of probability measures in reproducing kernel Hilbert spaces provide a flexible framework for fully nonparametric hypothesis tests of independence, which can capture any type of departure from independence, including nonlinear associations and multivariate interactions. However, these approaches come wi…
Paper proposes data quality measures for large-scale high-dimensional data.
problem Lack of practical data quality measures for large-scale high-dimensional data.
method Proposes two data quality measures: class separability and in-class variability. Efficient algorithms based on random projections and bootstrapping are provided.
result Efficient algorithms for computing data quality measures on large-scale high-dimensional data.
SignSGD outperforms SGD in linear regression with optimal scaling laws under PLRF model.
problem Improving linear regression performance with signSGD under power-law random features.
method Analysis of signSGD risk under PLRF model, comparison with SGD, identification of unique effects.
result SignSGD can have a steeper compute-optimal slope than SGD in noisy regimes, especially with WSD schedule.
Nonlinear dimensionality reduction embeddings computed from datasets do not provide a mechanism to compute the inverse map. In this paper, we address the problem of computing a stable inverse map to such a general bi-Lipschitz map. Our approach relies on radial basis functions (RBFs) to interpolate the inverse map ever…
The study examines how extra compute during testing affects the performance of large language models.
problem Understanding the conditions under which test-time scaling improves model performance.
method An in-context weight prediction task for linear regression was used to train transformers. The performance was analyzed under varying levels of test-time compute.
result Training transformers on diverse, relevant, and hard tasks leads to the best performance for test-time scaling.
New method speeds up learning of complex dynamical systems.
problem Efficiently learning large-scale dynamical systems from finite data.
method Random projections (sketching) to boost kernel-based Koopman operator estimators.
result The proposed estimators maintain accuracy while significantly reducing computation time.
Researchers parallelize neural kernels for large-scale data, achieving state-of-the-art accuracy.
problem Limited scalability of neural kernels on large datasets.
method Massively parallel computation across many GPUs, combined with a distributed, preconditioned conjugate gradients algorithm.
result Achieved state-of-the-art accuracy of 91.2% on CIFAR-5m dataset using neural kernels.
A new TwinGP framework for efficient large-scale GP modeling.
problem Efficiently modeling large-scale Gaussian processes with computational constraints.
method Combines global and local approximations using a subset-of-data approach.
result TwinGP framework performs on par or better than state-of-the-art methods at a fraction of the computational cost.