Deep dive: model merging

*** Part 2 is now available at • Deep dive: model mergi... : Model Breadcrumbs, Model Stock, DELLA
Model merging is an increasingly popular technique that makes it possible to add or remove capabilities to transformer models, without the need for any additional training.
In this video, we first introduce what model merging is. Then, we discuss different merging algorithms implemented in the mergekit library (github.com/arcee-ai): model soups, SLERP, Task Arithmetic, TIES, DARE, and Franken-merging.
⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos. Follow me on Medium at / julsimon or Substack at julsimon.substack.com. ⭐️⭐️⭐️
01:16 What is model merging?
07:10 Model soups
14:00 Spherical Linear Interpolation (SLERP)
20:35 Task Arithmetic
27:15 Trim, Extract Sign and Merge (TIES)
36:20 Drop and Rescale (DARE)
43:40 Franken-merging

[@kenchang3456]

Thank you for this video. I gotta give this a try 🙂

[@juliensimonfr]

You're welcome, and yes, you should :)

[@melikanobakhtian6018]

That was great and it helped me so much! Is there this possibility to have the presentation slides?

[@gnibu42]

Super intersting Julien, thanks a lot for sharing

[@juliensimonfr]

Glad you enjoyed it

[@SrikanthIyer]

Thanks for the fantastic video. Loved how you simplified almost all the methods to merge the models!

[@juliensimonfr]

Glad it was helpful!

[@subhamkundu5043]

Hey @Julien, great vide. I have a question regarding the scale factor in TIES method. How we determine the scale factor?

[@juliensimonfr]

Thank you. It's up to you, depending on how much you want to "influence" the base model. mergekit has a parameter called 'density': fraction of weights in differences from the base model to retain. Example at github.com/arcee-ai/mergekit/blob/edd3817e4a470c7a959ef4c505f52a650a46ff07/examples/ties.yml

[@uygarkurtai]

Great viedo thank you! What I didn't grasp quite well is that, let's say I'm merging 2 models. One is trained on maths, other is trained on coding. Do we expect the merged model to perform high level in both tasks?

[@juliensimonfr]

Yes, that's the expectation :)

[@abse-mj8pw]

I can't help wondering if there is an experiment which really fully discovers those technique like applying to all kinds of models or combining different methods together?

[@juliensimonfr]

Check out arcee.ai, their platform is definitely going that way.

[@abse-mj8pw]

@@juliensimonfr Thanks for your answer!! I've found some interesting blogs about it!

[@AbdennacerAyeb]

This is a random comment to boost your channel. Thank you.

[@juliensimonfr]

LOL, thank you.