Data

Tools

Indexes

Applications

Experiments

Open Source Science

lcc

Fine-tuning classifier NNs with Latent Cluster Correction

https://github.com/altaris/lcc

Science Score: 67.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ✓
    CITATION.cff file
    Found CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ✓
    .zenodo.json file
    Found .zenodo.json file
  • ✓
    DOI references
    Found 4 DOI reference(s) in README
  • ✓
    Academic publication links
    Links to: arxiv.org, zenodo.org
  • ○
    Committers with academic emails
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (10.6%) to scientific vocabulary

Keywords

fine-tuning huggingface latent-space pytorch-lightning torch torchvision
Last synced: 6 months ago · JSON representation ·

Repository

Fine-tuning classifier NNs with Latent Cluster Correction

Basic Info
  • Host: GitHub
  • Owner: altaris
  • License: mit
  • Language: Jupyter Notebook
  • Default Branch: master
  • Homepage:
  • Size: 17.6 MB
Statistics
  • Stars: 0
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Topics
fine-tuning huggingface latent-space pytorch-lightning torch torchvision
Created over 2 years ago · Last pushed 12 months ago
Metadata Files
Readme Changelog Contributing License Citation

README.md

LCC: Latent Cluster Correction

Python 3.10 CUDA 12 Documentation License DOI Paper

  • Neural networks take input samples and transform them into latent representations
  • Semantically similar samples tend to aggregate into latent clusters
  • This repository implements Latent Cluster Correction, a new technique to improve said latent clusters

Pretty images

These are examples of input datasets fed into image classifier models. Some selected latent representations are extracted and plotted in 2D (via dimensionality reduction). Initially, during the feature extraction phase, the samples are not clearly separated. But as the samples progressively get into the classification phase, visible latent clusters emerge. The goal of LCC is to help the formation of these clusters.

Installation

Make sure uv is installed. Then run

sh uv python install 3.10 uv sync --all-extras

Usage

  • Fine-tuning with LCC: modify and run lcc.sh, or use the CLI directly:

sh uv run python -m lcc train --help

For example:

sh uv run python -m lcc train \ microsoft/resnet-18 \ PRESET:cifar100 \ output_dir \ --batch-size 256 \ --head-name classifier.1 \ --logit-key logits \ --lcc-submodules resnet.encoder.stages.3 \ --lcc-warmup 1 \ --lcc-weight 0.01 \ --seed 123

  • Pretty-print a model structure from HuggingFace: run ./pretty-print.sh HF_MODEL_NAME, e.g.

sh ./pretty-print.sh microsoft/resnet-18

API overview

Cite

bibtex @misc{hothanhImprovingFineTuningLatent2025, title = {Improving {{Fine-Tuning}} with {{Latent Cluster Correction}}}, author = {Ho Thanh, C{\'e}dric}, year = {2025}, month = jan, number = {arXiv:2501.11919}, eprint = {2501.11919}, primaryclass = {cs}, publisher = {arXiv}, doi = {10.48550/arXiv.2501.11919}, urldate = {2025-01-22}, archiveprefix = {arXiv}, keywords = {Computer Science - Machine Learning}, }

  • Code

bibtex @software{Ho_Thanh_LCC_Latent_Cluster_2025, author = {Ho Thanh, Cédric}, license = {MIT}, month = jan, title = {{LCC: Latent Cluster Correction}}, url = {https://github.com/altaris/lcc}, version = {1.0.0}, year = {2025} }

Owner

  • Name: Cédric
  • Login: altaris
  • Kind: user
  • Location: Japan
  • Company: RIKEN

Citation (CITATION.cff) 

# This CITATION.cff file was generated with cffinit.
# Visit https://bit.ly/cffinit to generate yours today!

cff-version: 1.2.0
title: "LCC: Latent Cluster Correction"
message: "If you use this software, please cite it as below."
type: software
authors:
  - family-names: Ho Thanh
    given-names: Cédric
    orcid: "https://orcid.org/0000-0003-4476-2034"
identifiers:
  - type: doi
    value: 10.5281/zenodo.14934702
repository-code: "https://github.com/altaris/lcc"
url: "https://github.com/altaris/lcc"
license: MIT
commit: 880b2b16bf3ca39f77e19be4550db57a4d3ae79f
version: 1.0.0
date-released: "2025-01-27"

GitHub Events

Total
  • Delete event: 6
  • Public event: 1
  • Push event: 27
  • Create event: 7
Last Year
  • Delete event: 6
  • Public event: 1
  • Push event: 27
  • Create event: 7

Committers

Last synced: 7 months ago

All Time
  • Total Commits: 498
  • Total Committers: 1
  • Avg Commits per committer: 498.0
  • Development Distribution Score (DDS): 0.0
Past Year
  • Commits: 259
  • Committers: 1
  • Avg Commits per committer: 259.0
  • Development Distribution Score (DDS): 0.0
Top Committers
Name Email Commits
Cédric HT a****s 498

Issues and Pull Requests

Last synced: 7 months ago

All Time
  • Total issues: 0
  • Total pull requests: 0
  • Average time to close issues: N/A
  • Average time to close pull requests: N/A
  • Total issue authors: 0
  • Total pull request authors: 0
  • Average comments per issue: 0
  • Average comments per pull request: 0
  • Merged pull requests: 0
  • Bot issues: 0
  • Bot pull requests: 0
Past Year
  • Issues: 0
  • Pull requests: 0
  • Average time to close issues: N/A
  • Average time to close pull requests: N/A
  • Issue authors: 0
  • Pull request authors: 0
  • Average comments per issue: 0
  • Average comments per pull request: 0
  • Merged pull requests: 0
  • Bot issues: 0
  • Bot pull requests: 0
View more stats
Top Authors
Issue Authors
Pull Request Authors
Top Labels
Issue Labels
Pull Request Labels

Dependencies

.github/workflows/gh-pages.yml actions
  • actions/checkout v3 composite
  • actions/setup-python v3 composite
  • peaceiris/actions-gh-pages v3 composite
.etc/setup.py pypi
pyproject.toml pypi
  • click >=8.1.7
  • datasets [vision]>=3.0.0
  • faiss-cpu >=1.9.0
  • loguru >=0.7.2
  • more-itertools >=10.5.0
  • networkx >=3.3
  • numpy >=1.25
  • pillow >=10.4.0
  • pytorch-lightning >=2.4.0
  • regex >=2024.9.11
  • safetensors >=0.4.5
  • scikit-learn >=1.5.2
  • tensorboard >=2.17.1
  • timm >=1.0.9
  • torch >=2.4.1
  • transformers >=4.44.2
  • turbo-broccoli >=4.12.2
requirements.nocuda.txt pypi
  • faiss *
  • umap-learn *