I am at the International Conference of Machine Learning (ICML) and will be presenting our work on Protein Language Models. If you are there and want to meet and talk about PLMs or NLP, DM me.
بندور على MLOps Engineer في التيم بتاعنا في Proteinea. حابين حد ييجي يتعلم الrole عندنا و يكون عنده 1-3 سنين خبرة كdevops or backend software engineer. بص على الrequirements و لو لقيت 50% أو أكتر منها عندك، قدم عندنا. نحب حد بيحب يتعلم حاجات جديدة و عنده مهارات تواصل كويسة.
Ankh was trained on @Google 's TPUv4 Pods. Thanks to @googlecloud 's generous fund, we are able to release the code and model weights free for academic and noncommercial use cases.
code: https://t.co/NWQQ8kAG7y
model weights on @huggingface :https://t.co/4svrPsyCjk
(3/n)
Ankh was tested heavily on various principal structure and function benchmarks for protein engineering, from fitness to contact prediction, achieving SoTA performance on them!
It can also produce a biologically meaningful library of proteins -- from just a single sequence.
(2/n)
Introducing Ankh ☥ :
An optimized protein language model (PLM) achieving SoTA performance with < 10% of the top/closest-performing models’ parameters. It is benchmarked on various downstream protein tasks, and tested for protein generation.
https://t.co/7VYwz748TV
(1/n)
As of now, Proteinea has an official physical existence in KAUST as part of the Destination Deep Tech program (we even have our own offices ! come say hi to us in the entrepreneurship center at @KAUSTinnovation cluster).
We took the KAUST lab tour, and were mesmerized by the technological beauties we witnessed
We are looking forward to meeting Saudi companies in the field, and developing our IP in collaboration with KAUST faculty
@TimSDobbs@aryelipman It was one year ago our R&D team came up with such an idea. We similarly googled it, read some papers, and that pivoted us from insect farming into full-stack protein production, and only a few months ago we graduated @indbio !