Research
On-device research index

arXiv research

A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.

168,695 papers · 148 categories

Trend · papers per month

591418 · May 202619922001200920172026
48 results for Checkpoint Merging

Signed-permutation coordinate transport improves model alignment across checkpoints.

problem Improper alignment of coordinate-indexed objects across model checkpoints.
method Introduces sign-marginalized Hungarian matching and coordinate-preserving transport.
result Recovering signed-permutation gauge improves coordinate alignment and model performance.

ScheduleFree+ improves large language model training without schedules or learning rates.

problem Scaling up Schedule-Free Learning to large language models.
method Learning-rate-free and schedule-free method for training large language models.
result ScheduleFree+ outperforms SOTA schedules by 31% at 1000 tokens per parameter.

Study shows training duration impacts model merging quality, suggesting joint selection of duration and method.

problem Impact of expert training duration on model merging quality for large language models (LLMs).
method Systematically fine-tuned experts on five domains across three model sizes, evaluating five merging methods at each duration.
result Training duration affects merging quality, with simple averaging degrading sharply and sparsification-based methods performing well past the validation optimum.

Study shows training duration affects model merging quality, suggesting joint selection of duration and method.

problem Impact of expert training duration on model merging quality for large language models (LLMs).
method Systematically fine-tuned experts on five domains across three model sizes, evaluated five merging methods at each duration.
result Training duration and merging method should be chosen jointly, not independently.

The paper tackles model selection for unseen tasks by capturing relationships among checkpoints.

problem Deciding which model combinations are likely to be effective for a new task is difficult.
method The paper models the task space as a Gaussian process and identifies representative checkpoints using mutual information and a greedy algorithm.
result Representative checkpoints generalize to new tasks with superior performance.

TopoGeoScore selects robust checkpoints using only source-domain representations.

problem Selecting robust checkpoints without target-domain labels or samples.
method Constructs class-conditional mutual k-nearest-neighbour graphs and extracts three interpretable signals.
result Source representations contain measurable global-local-topological evidence of robustness.

SEP uses checkpoints to protect data from training good models.

problem Protecting data from competitors training high-performance models.
method Forming perturbed examples using model checkpoints' gradients to ensure they are always unrecognized.
result SEP significantly reduces model accuracy when trained on perturbed data, demonstrating its effectiveness.

ACA method improves gradient estimation for neural ODEs, reducing error and training time.

problem Inaccurate gradient estimation methods hinder the performance of neural ODEs on benchmark tasks.
method Adaptive Checkpoint Adjoint (ACA) method that applies trajectory checkpointing, deletes redundant components, and supports adaptive solvers.
result ACA reduces error rate by half and training time by half compared to adjoint and naive methods on image classification tasks.

New ODE solvers improve training efficiency and accuracy.

problem Training Neural ODEs requires efficient and accurate gradient calculation.
method Presented algebraically reversible ODE solvers that are time and memory efficient, calculate exact gradients, and are numerically stable.
result Reversible solvers strictly improve upon previous architectures in efficiency and accuracy.

LEWIS merges LLMs without training, improving performance on specific tasks.

problem Limited performance improvement of merged models on specific benchmarks.
method Guided model merging using layer-wise sparsity and task-vector pruning.
result Improved model performance by up to 11.3% on math-solving tasks.

Deep Neural Networks(DNNs) require huge GPU memory when training on modern image/video databases. Unfortunately, the GPU memory is physically finite, which limits the image resolutions and batch sizes that could be used in training for better DNN performance. Unlike solutions that require physically upgrade GPUs, the G…

2018-07-31abs ↗pdf ↗

Machine learning (ML) training algorithms often possess an inherent self-correcting behavior due to their iterative-convergent nature. Recent systems exploit this property to achieve adaptability and efficiency in unreliable computing environments by relaxing the consistency of execution and allowing calculation errors…

2018-10-17abs ↗pdf ↗

EpiMer merges models by solving Fréchet mean on a Riemannian manifold.

problem Integrating knowledge from multiple models without retraining.
method EpiMer casts model merging as solving the Fréchet mean on a Riemannian manifold, restricting computation to a low-rank subspace.
result EpiMer outperforms flat-geometry methods on image classification tasks.

Since DeepMind's AlphaZero, Zero learning quickly became the state-of-the-art method for many board games. It can be improved using a fully convolutional structure (no fully connected layer). Using such an architecture plus global pooling, we can create bots independent of the board size. The training can be made more …

2020-01-27abs ↗pdf ↗

Securely evaluates the benefits of merging datasets for causal estimation.

problem Challenges in assessing the value of merging datasets for causal treatment effect estimation.
method Cryptographically secure multi-party computation to evaluate Expected Information Gain (EIG) while ensuring privacy.
result Demonstrates the first privacy-preserving method for dataset acquisition tailored to causal estimation.

A new method merges neural networks using CCA to improve model performance.

problem Improving model accuracy through ensembling while reducing computational and storage costs.
method CCA Merge, a new model merging algorithm based on Canonical Correlation Analysis.
result CCA Merge leads to better model performance than past methods, especially in merging more than two models.

This paper reverses a construction by merging boundary critical points into an interior one.

problem Pushing interior critical points to the boundary and splitting them into two boundary points.
method Specific assumptions allow merging two boundary critical points into one interior critical point.
result Merging two boundary critical points into a single interior critical point.

NAMEx merges experts using Nash bargaining for improved performance.

problem Sparse Mixture of Experts merging strategies lack a principled weighting mechanism.
method Reinterpreting expert merging through game theory, introducing Nash Merging and complex momentum.
result NAMEx consistently outperforms competing methods across various tasks and system sizes.

Discrete knot theory models use lattice-filtered graphs to detect merging knot components.

problem Detecting merging knot components in discrete models.
method Lattice-filtered move graphs to model knot types, identifying connected components and merge scales.
result Merge scale defined by connected components of lattice-filtered move graphs, with specific examples for the figure-eight knot.

Method constructs finance LLMs without instruction data using pretraining and model merging.

problem Developing domain-specific LLMs for finance is resource-intensive.
method Continual pretraining on financial data + model merging of instruction-tuned and domain-specific pretrained vectors.
result Successfully constructs instruction-tuned LLMs for finance without additional instruction data.

Extracting a curriculum from a teacher network improves distillation efficiency.

problem Efficiently training a small network using a large teacher network's output.
method Random projection of teacher network's hidden representations to progressively train the student network.
result Extracted curriculum significantly outperforms one-shot distillation and achieves similar performance to progressive distillation.

Deep learning algorithms for connectomics rely upon localized classification, rather than overall morphology. This leads to a high incidence of erroneously merged objects. Humans, by contrast, can easily detect such errors by acquiring intuition for the correct morphology of objects. Biological neurons have complicated…

2017-05-30abs ↗pdf ↗

DPSM clusters nodes in data and graph spaces via density propagation and subcluster merging.

problem Automatic clustering of nodes in data and graph spaces.
method Density-based node clustering with propagation process and spectral clustering on subclusters.
result DPSM effectively clusters nodes in both data and graph spaces.

Bayesian Federated Inference improves survival model analysis without merging data.

problem Accurately estimating survival model parameters requires sufficient data and events, which is often lacking in practice.
method Bayesian Federated Inference (BFI) for survival models, where local centers perform analyses and combine results.
result Results from BFI are similar to those from merged data analyses, demonstrating excellent performance.

The hierarchical Dirichlet process (HDP) has become an important Bayesian nonparametric model for grouped data, such as document collections. The HDP is used to construct a flexible mixed-membership model where the number of components is determined by the data. As for most Bayesian nonparametric models, exact posterio…

2012-01-08abs ↗pdf ↗

NetFuse merges different DNN models with varying weights for faster inference.

problem Inference speed of DNN models with different weights cannot be improved using existing techniques.
method NetFuse merges models with the same architecture but different weights and inputs, replacing operations with more general ones.
result NetFuse can speed up DNN inference time up to 3.6x on a NVIDIA V100 GPU.