legacy-datasets/wikipedia
Updated • 122k • 632
How to use GeneZC/bert-base-minimoe-4L-384H with Transformers:
# Load model directly
from transformers import AutoTokenizer, MoEBertMiniLM
tokenizer = AutoTokenizer.from_pretrained("GeneZC/bert-base-minimoe-4L-384H")
model = MoEBertMiniLM.from_pretrained("GeneZC/bert-base-minimoe-4L-384H")minimoe-4L-384H distilled from bert-base-uncased on Wikipedia.
Repository: https://github.com/GeneZC/MiniMoE arXiv: https://arxiv.org/abs/2305.12129