Predicting integers from continuous parameters

Hugging Face Daily Papers Papers

Summary

Study proposes Bitwise and discrete Laplace distributions to predict integer labels directly from continuous neural-network parameters, outperforming standard regression across tabular, sequential and image tasks.

We study the problem of predicting numeric labels that are constrained to the integers or to a subrange of the integers. For example, the number of up-votes on social media posts, or the number of bicycles available at a public rental station. While it is possible to model these as continuous values, and to apply traditional regression, this approach changes the underlying distribution on the labels from discrete to continuous. Discrete distributions have certain benefits, which leads us to the question whether such integer labels can be modeled directly by a discrete distribution, whose parameters are predicted from the features of a given instance. Moreover, we focus on the use case of output distributions of neural networks, which adds the requirement that the parameters of the distribution be continuous so that backpropagation and gradient descent may be used to learn the weights of the network. We investigate several options for such distributions, some existing and some novel, and test them on a range of tasks, including tabular learning, sequential prediction and image generation. We find that overall the best performance comes from two distributions: Bitwise, which represents the target integer in bits and places a Bernoulli distribution on each, and a discrete analogue of the Laplace distribution, which uses a distribution with exponentially decaying tails around a continuous mean.
Original Article
View Cached Full Text

Cached at: 04/22/26, 06:17 AM

Paper page - Predicting integers from continuous parameters

Source: https://huggingface.co/papers/2602.10751

Abstract

Research examines direct modeling of integer-labeled data using discrete probability distributions with continuous parameters suitable for neural network training, evaluating Bitwise and discrete Laplace distributions against traditional regression approaches.

We study the problem of predicting numeric labels that are constrained to the integers or to a subrange of the integers. For example, the number of up-votes on social media posts, or the number of bicycles available at a public rental station. While it is possible to model these as continuous values, and to apply traditional regression, this approach changes the underlying distribution on the labels from discrete to continuous.Discrete distributionshave certain benefits, which leads us to the question whether suchinteger labelscan be modeled directly by a discrete distribution, whose parameters are predicted from the features of a given instance. Moreover, we focus on the use case of output distributions ofneural networks, which adds the requirement that the parameters of the distribution be continuous so thatbackpropagationandgradient descentmay be used to learn the weights of the network. We investigate several options for such distributions, some existing and some novel, and test them on a range of tasks, includingtabular learning,sequential predictionandimage generation. We find that overall the best performance comes from two distributions: Bitwise, which represents the target integer in bits and places a Bernoulli distribution on each, and a discrete analogue of the Laplace distribution, which uses a distribution with exponentially decaying tails around a continuous mean.

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2602\.10751

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2602.10751 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2602.10751 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2602.10751 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Variable Bit-width Quantization: Learning Per-Group Precision for "Bigger-but-Smaller" Language Models

arXiv cs.LG

Introduces Variable Bit-width Quantization (VBQ), a training-time method where each group of 64 weights learns its own bit-width (1,2,4,8) via Gumbel-Softmax relaxation. VBQ discovers a heterogeneous allocation that yields a 'bigger-but-smaller' regime, e.g., a 131M parameter model at 1.82 mean bits beats a 55M FP16 model while using less storage, and a 1.46B model matches a 593M FP16 with ~3.7x less storage.