metadata
language:
- en
tags:
- biology
- ESM
- language-model
- protein
- VEP
pretty_name: VESM scores
size_categories:
- 100M<n<1B
Training and Inference Datasets for VESM
This repository provides datasets for training and evaluating VESM models in the paper "Compressing the collective knowledge of ESM into a single protein language model" by Tuan Dinh, Seon-Kyeong Jang, Noah Zaitlen and Vasilis Ntranos.
- train.zip: training datasets
- benchmarks.zip: evaluation datasets
- base_llrs.zip: individual LLRS used for creating ESMIN dataset.
Please see the corresponding GitHub repo (https://github.com/ntranoslab/vesm) for more details.
License
The predictions of VESM_3B are distributed under the MIT License.