Imageomics / Imageomics/pybioclip

How to address updating original BioCLIP Model

Open
#160 0 comments 0 reactions 0 assignees View on GitHub
question
Dominant language
Python
Stars
67
Forks
14
PR merge metrics
No merged PRs in 30d

Description

We're going to update the original [BioCLIP model](https://huggingface.co/imageomics/bioclip) to use a corrected taxonomy (retrained with [TreeOfLife-10M](https://huggingface.co/datasets/imageomics/TreeOfLife-10M), just with the taxonomy aligned through TaxonoPy). Updating the model weights on the existing repository will change the results of pybioclip for earlier versions (see the discussion in #111, where we discussed challenges in using a particular revision, limits of our settled approach are shown in #113).

We could optionally release a patch for the v1 series (where original BioCLIP is the default) that allows a workaround to at least report the model version.

Contributor guide

No contributing guide indexed for this repository

Research direction

Start by reading the discussion in #111 and the limits described in #113, then inspect how the v1 series selects the original BioCLIP model and reports its version. Done means reaching agreement on the model-update or patch approach and defining how earlier results and model versions will be identified.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.