Merging Multi-Task Models via Weight-Emsembling Mixture of Experts

I reimplemented this using the CLIP models from HuggingFace at https://github.com/tanganke/fusion_bench

Experimental Results

The results of the experiments in the paper are placed at results.

Some of the code in this repository is based on the following repositories:

Name		Name	Last commit message	Last commit date
Latest commit History 13 Commits
config		config
figure		figure
results		results
scripts		scripts
src		src
.gitignore		.gitignore
README.md		README.md
offline_mode.sh		offline_mode.sh
plot_results.ipynb		plot_results.ipynb
pyproject.toml		pyproject.toml
requirements.txt		requirements.txt
routing_analysis.ipynb		routing_analysis.ipynb