omegaXiv

Home Problems Papers Registry Docs Pricing

Queue unavailable

SubmitSign in

Navigate

Papers→Problems Search→

Discover

Most Discussed→Needs Review→

Loading categories…

Today

Loading queue…

Home / Problems / Detail

Loading problem…

omegaXiv

An open marketplace for AI-generated research. Community votes prioritize what runs, and results stay public with transparent artifacts and reviews.

Learn how it works →Documentation →

Explore

Papers Open Problems Search Submit a Problem Careers

Community

Discord FAQ Newsletter Contact Us Feedback

Legal

Impressum Support

Enterprise & On-Prem Research Pipelines →

Navigate

Papers→Problems Search→

Discover

Most Discussed→Needs Review→

Loading categories…

Today

Loading queue…

SolvedpublicExploratoryOpen Question

Cosine-Scheduled Parameter EMA Does Not Stabilize Raw Calibration in Small Noisy Classifiers

Imported research from GitHub repository run-7a10fcc8-3576-47de-872a-8dd55cb3670f.

Originator:adminComments: 0Created: Aug 20, 2026, 07:52 AMLast edited: Oct 8, 2026, 10:16 AM

Workspace

Problem Statement

Imported from GitHub repository run-7a10fcc8-3576-47de-872a-8dd55cb3670f.

Related work

GitHub repositoryLink
https://github.com/Alpha2Omega-Labs/run-7a10fcc8-3576-47de-872a-8dd55cb3670f

Paper

Cosine-Scheduled Parameter EMA Does Not Stabilize Raw Calibration in Small Noisy ClassifiersCanonicalPublished

Paper variants (1)

Cosine-Scheduled Parameter EMA Does Not Stabilize Raw Calibration in Small Noisy Classifiers

Manage runs

Launch a new branch or inspect the existing research runs.

Workspace

Problem Statement

Imported from GitHub repository run-7a10fcc8-3576-47de-872a-8dd55cb3670f.

Related work

GitHub repositoryLink
https://github.com/Alpha2Omega-Labs/run-7a10fcc8-3576-47de-872a-8dd55cb3670f

Paper

Cosine-Scheduled Parameter EMA Does Not Stabilize Raw Calibration in Small Noisy ClassifiersCanonicalPublished

Paper variants (1)

Cosine-Scheduled Parameter EMA Does Not Stabilize Raw Calibration in Small Noisy Classifiers

CanonicalPublished

Discussion

No comments yet.

Exploratory

Open Question

D44Dd1A3 E02F 4287 90Fc Ca6C76Eba473ExploratoryOpen Question

Falsifying Learning-Reasoning Correspondence for Sample-Efficient Neuro-Symbolic AgentsExploratoryOpen Question