Join the conversation

Join the community of Machine Learners and AI enthusiasts.

Sign Up
thomwolf 
posted an update Apr 6
Post
3123
Very interesting model just released by MyShell: jetmoe/jetmoe-8b . It's a 8B-parameters MoE LLM so 2.2B active parameters, really efficient.

Main characteristics:
- impressive performances for its size (beating meta-llama/Llama-2-7b and huggyllama/llama-13b)
- combine Mixture of Attention heads (MoA) and Mixture of MLP Experts (MoE) – 8 experts with 2 being active for each token
- trained on a rather limited 1.25T tokens from publicly available datasets – training recipe follows the MiniCPM's two-phases training method => first time I see this for a 2B+ model
- $100k to train
- open weights - open sharing of recipes - open dataset - open code => ♡
- still interesting room to improve performances (be it only by training longer)

Links:
- report: https://research.myshell.ai/jetmoe
- model: jetmoe/jetmoe-8b
- code: https://github.com/myshell-ai/JetMoE

Note: I actually detailed all of the MiniCPM schedule, Mixture-of-expert (MoE) and many of the datasets used in this work in my recent little guide to building LLMs in 2024, so feel free to check it out if you want to learn more on these topics: https://www.youtube.com/watch?v=2-SPH9hIKT8

@thomwolf very good development indeed! I thought it is done by some individual/s but when I see all the big names & they just spent $100k then I am disappointed. Combined they should have done more and they can spend more at least on compute.