Adaptive batch size schedules improve language model training efficiency and generalization.
problem Dilemma of choosing batch sizes in large-scale model training.
method General-purpose adaptive batch size schedules compatible with data and model parallelism.
result Adaptive batch size schedules outperform constant batch sizes and heuristic warmup schedules.
Convolutional neural networks (CNNs) are commonly trained using a fixed spatial image size predetermined for a given model. Although trained on images of aspecific size, it is well established that CNNs can be used to evaluate a wide range of image sizes at test time, by adjusting the size of intermediate feature maps.…
We determine the critical batch size for large language models and find it scales with data size, not model size.
problem Determining the optimal batch size for large-scale model training.
method We propose a measure of critical batch size, pre-trained models, and systematic hyper-parameter sweeps.
result The critical batch size scales primarily with data size, not model size.
This work studies scaling laws for low-precision training in high-dimensional linear regression.
problem Optimizing trade-off between model quality and training costs in high-dimensional linear regression.
method Theoretical study of scaling laws for low-precision training within a high-dimensional sketched linear regression framework, analyzing multiplicative and additive quantization.
result Multiplicative quantization maintains full-precision model size, while additive quantization reduces effective model size.
A new scaling law predicts optimal batch size for training models.
problem Finding the optimal batch size for training models efficiently.
method Proposed a three-term scaling law that considers model size, training data, training steps, and batch size.
result The three-term law accurately recovers the optimal batch size and can be robustly fit with fewer training runs.
Study shows model size inversely affects few-shot instruction accuracy.
problem Understanding how model size impacts few-shot instruction prompting accuracy.
method Introduced DeltaWords dataset to evaluate model's ability to follow instructions.
result Model size inversely affects few-shot instruction accuracy, with larger models performing worse.
Revisits granular models explaining firm growth rates and sizes.
problem Understanding the relationship between firm size and growth rate statistics.
method Developed new theoretical insights linking firm size and growth rate statistics within granular models.
result Growth volatility distribution is size-independent but fat-tailed, challenging granular models.
Paper presents a new way to estimate model changes without full model evaluation.
problem Efficiently estimating changes in model parameters and outputs due to data point removal.
method Dual representation of influence functions for linearizable models, reducing computational complexity.
result The dual representation can be an efficient alternative to original influence functions, especially for large models.
Paper reduces recommender system model size by 90%.
problem Large model size in recommender systems.
method Hybrid frequency and double hashing for model size reduction.
result 90% reduction in model size with no performance loss.
Scaling laws for neural language models reveal optimal model size and compute allocation.
problem Understanding the optimal model size and compute allocation for neural language models.
method Empirical analysis of scaling laws for cross-entropy loss across model size, dataset size, and compute.
result Simple equations govern the dependence of overfitting and training speed on model/dataset size and model size, respectively.
Most generative models for clustering implicitly assume that the number of data points in each cluster grows linearly with the total number of data points. Finite mixture models, Dirichlet process mixture models, and Pitman--Yor process mixture models make this assumption, as do all other infinitely exchangeable cluste…
Automatically adjusts model size for continual Gaussian processes.
problem Determining optimal model size for incremental data.
method Automatic adjustment of model size during continual learning.
result Near-optimal performance across various datasets without hyperparameter tuning.
A simple block configures optimal kernel sizes for time series classification.
problem Choosing the right kernel size for time series classification.
method Proposes Omni-Scale block (OS-block) with kernel sizes determined by prime numbers.
result Models with OS-block achieve state-of-the-art performance on time series benchmarks.
Polyak step size GD reaches final radius of convergence after log iterations.
problem Statistical and computational complexities of Polyak step size GD.
method Generalized smoothness and Lojasiewicz conditions, stability of gradients.
result Polyak step size GD reaches final statistical radius of convergence after logarithmic number of iterations.
Support vector regression (SVR) has been widely used to reduce the high computational cost of computer simulation. SVR assumes the input parameters have equal sample sizes, but unequal sample sizes are often encountered in engineering practices. To solve this issue, a new prediction approach based on SVR, namely as hig…
AdAdaGrad optimizes batch sizes for deep learning models, reducing the generalization gap.
problem The generalization gap between large-batch and small-batch training in deep learning.
method AdAdaGrad introduces adaptive batch size strategies derived from adaptive sampling methods.
result AdAdaGradNorm converges to a first-order stationary point with a rate of O(1/K) in K iterations.
Ensembles of random-feature models can't outperform a single large model.
problem Finding the optimal balance between model size and ensemble size.
method Deterministic equivalent risk estimates and scaling laws analysis.
result Ensembles of random-feature models achieve near-optimal performance only under specific conditions.
A microscopic model of aggregation and fragmentation is introduced to investigate the size distribution of businesses. In the model, businesses are constrained to comply with the market price, as expected by the customers, while customers can only buy at the prices offered by the businesses. We show numerically and ana…
We address the issue of the distribution of firm size. To this end we propose a model of firms in a closed, conserved economy populated with zero-intelligence agents who continuously move from one firm to another. We then analyze the size distribution and related statistics obtained from the model. Our ultimate goal is…
This paper compares Transformers and RNNs in various tasks, showing size differences.
problem Comparing representational capabilities of Transformers and RNNs across tasks.
method Analysis of differences in tasks like index lookup, nearest neighbor, and string equality.
result Size differences in Transformers and RNNs for various tasks.
Seesaw optimizes training by balancing learning rate and batch size, accelerating model pretraining.
problem Optimizing training efficiency for large language models with adaptive optimizers.
method Develops a principled framework for batch-size scheduling, introducing Seesaw which multiplies learning rate by 1/√2 and doubles batch size.
result Empirically, Seesaw reduces wall-clock time by approximately 36% compared to cosine decay, matching theoretical limits.
Optimal microlending group size is 5 people.
problem Determining the best group size for microlending to minimize default risk.
method Mathematical modeling with interacting forces and precise hypotheses.
result The optimal microlending group size is 5 people.
Empirical study on SGD hyperparameters and adversarial robustness.
problem Effect of SGD hyperparameters on adversarial robustness and generalization.
method Empirical observation of learning rate, batch size, and momentum effects on adversarial robustness and generalization.
result Constant learning rate to batch size ratio leads to good generalization and almost constant adversarial robustness.
Value selection reduces model size while maintaining accuracy.
problem Space efficiency in model size reduction.
method Two probabilistic methods based on information theory's metric: PVS and P + VS.
result Value selection achieves balance between accuracy and model size reduction.
Adaptive batch sizes improve local gradient methods in distributed training.
problem Communication bottlenecks in distributed deep learning.
method Adaptive batch size strategies for local gradient methods.
result Adaptive batch sizes reduce minibatch gradient variance and improve training efficiency.
We introduce a new system of stochastic differential equations which models dependence of market beta and unsystematic risk upon size, measured by market capitalization. We fit our model using size deciles data from Kenneth French's data library. This model is somewhat similar to generalized volatility-stabilized model…
As neural networks become widely deployed in different applications and on different hardware, it has become increasingly important to optimize inference time and model size along with model accuracy. Most current techniques optimize model size, model accuracy and inference time in different stages, resulting in subopt…
Pruning improves model generalization in over-parameterized models, contradicting traditional theories.
problem Pruning's effect on generalization in over-parameterized models.
method Empirical study on standard pruning algorithms and additional regularization effects.
result Pruning leads to better training and regularization, improving generalization.
This paper characterizes projective models in statistical relational learning.
problem Projectivity in statistical relational models is beneficial for inference and learning.
method Representation theorems for infinite exchangeable arrays to characterize projective models.
result A class of directed graphical latent variable models correspond to projective relational models.
pmsims R package uses Gaussian process for flexible sample size estimation in clinical models.
problem Determining adequate sample size for clinical prediction models.
method Simulation-based Gaussian process search for flexible sample size estimation.
result Gaussian process-based method produces more stable sample size estimates, especially in challenging settings.
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.
Proposes a differentiable hypergeometric distribution for learning group importance.
problem Learning the sizes of subsets in applications like clustering and weakly-supervised learning.
method Introduces a reparameterizable hypergeometric distribution to model group sizes and learn their relative importance.
result Outperforms previous methods in weakly-supervised learning and clustering.
In biospectroscopy, suitably annotated and statistically independent samples (e. g. patients, batches, etc.) for classifier training and testing are scarce and costly. Learning curves show the model performance as function of the training sample size and can help to determine the sample size needed to train good classi…
We introduce a simple agent-based model which allows us to analyze three stylized facts: a fat-tailed size distribution of companies, a `tent-shaped' growth rate distribution, the scaling relation of the growth rate variance with firm size, and the causality between them. This is achieved under the simple hypothesis th…
The relationship between the size and the variance of firm growth rates is known to follow an approximate power-law behavior σ(S)∼S−β(S) where S is the firm size and β(S)≈0.2 is an exponent weakly dependent on S. Here we show how a model of proportional growth which treats firms as classes compos…
Modeling functional data, this study uncovers the size-and-shape of functions under noisy observations.
problem Uncertainty in recovering a fixed effect function from noisy observations.
method Bayesian functional mixed model with priors on unitary transformations.
result It is possible to recover the size-and-shape of a square-integrable function μ. We add size factor to CAPM and normalize residuals by Volatility Index.
problem Capturing the size effect in CAPM and making residuals Gaussian.
method Insert size effect, normalize residuals by Volatility Index, and fit model to real-world data.
result The new model shows long-term stability and connects to Stochastic Portfolio Theory.
Compressing DNNs is important for the real-world applications operating on resource-constrained devices. However, we typically observe drastic performance deterioration when changing model size after training is completed. Therefore, retraining is required to resume the performance of the compressed models suitable for…
New neural network models learn symmetric functions of varying input sizes.
problem Learning symmetric functions with varying input sizes.
method Functional perspective on neural networks, treating symmetric functions as functions over probability measures.
result Established approximation and generalization bounds for shallow architectures that extend across input sizes.
New kernel model scales to large datasets.
problem Challenges in scaling kernel machines to large datasets.
method Introduced EigenPro 3.0, an algorithm based on projected dual preconditioned SGD.
result Ability to scale model and data sizes independently.
The present study introduce the human capital component to the Fama and French five-factor model proposing an equilibrium six-factor asset pricing model. The study employs an aggregate of four sets of portfolios mimicking size and industry with varying dimensions. The first set consists of three set of six portfolios e…
The paper examines sizing strategies for algorithmic trading in volatile markets.
problem High volatility creates challenges for algorithmic traders.
method Investigates different sizing models and backtesting techniques for financial trading.
result Sizing models can lower Value at Risk (VaR) during crisis events.
Proposes an exponentially increasing step-size for faster parameter estimation in statistical models.
problem Slow convergence of gradient descent in locally convex loss functions.
method Exponentially increasing step-size in gradient descent algorithm.
result Converges linearly to optimal solution under homogeneous assumptions.
Study finds that only a fraction of data is needed for accurate patient-level prediction models.
problem Developing predictive models for patient-level outcomes using large observational data.
method Empirical assessment of sample size effects on model performance and complexity using learning curves.
result A median reduction of 9.5% to 78.5% in the number of observations and 8.6% to 68.3% in the number of predictors can be achieved with adequate sample size.
Two methods reduce BN and DNN complexity, balancing size and accuracy.
problem Balancing model size and prediction accuracy in Bayesian networks and deep neural networks.
method Quantization-aware training and tree-augmented naive Bayes structure learning extension.
result Pareto optimal models found for small-scale scenarios.
This work provides a scaling rule for model EMA optimization across batch sizes.
problem Training dynamics and performance differences across batch sizes when using model EMA.
method Developed a scaling rule for model EMA optimization, demonstrating its validity across various architectures and data modalities.
result Enabled SSL methods like BYOL to train at larger batch sizes without performance degradation.
In an increasing number of domains it has been demonstrated that deep learning models can be trained using relatively large batch sizes without sacrificing data efficiency. However the limits of this massive data parallelism seem to differ from domain to domain, ranging from batches of tens of thousands in ImageNet to …
Deep model tackles claim size modeling with quantile-based regression.
problem Actuarial claim size modeling difficulty with no simple distribution.
method Deep composite regression model with quantile splicing point.
result Deep neural network regression models show superiority over classical approaches.