Hugging Face's logo

Spaces:
retrain-pipelines
/
README
Running

App Files Community
README / README.md
Aurelien-Morgan's picture
Aurelien-Morgan
Yet another revisit of the inspector short vid
25f2395 verified
preview code
|
raw
history blame contribute delete
2.22 kB
metadata
title: README
emoji: 📈
colorFrom: red
colorTo: yellow
sdk: static
pinned: false

Wondering if the new version of your model is better than the last ? Did continuous training improve, or decrease perf ? Does sorting through model cards feel like a never-ending chore ?

With retrain-pipelines, you can overlook your Hugging Face model versions, hassle-free :

with the inspector on the Hub
inspector.webm hub.webm

retrain-pipelines focuses on continuous learning (retraining through dataset updates/refinements), enabling the creation of a new version of a model and the comparison of its performance against that of its best predecessor.

Release v0.1.2 brings integration for the Hugging Face ecosystem. By retraining with retrain-pipelines, you benefit from :

  • automated dataset and model versioning, with dynamically-generated (metadata-rich) READMEs and full fine-grained lineage
  • automated upload of retraining source-code
  • automated upload of retraining pipeline-card
  • an inspector for model performance history across versions
  • and all other good things that make the retrain-pipelines' DNA, like highly-customizable pipeline templates to get started fast !