How to use thrunlab/Mistral_Sparse_pretraining_80_percent with Transformers:
# Load model directly from transformers import AutoTokenizer, SparseMistral tokenizer = AutoTokenizer.from_pretrained("thrunlab/Mistral_Sparse_pretraining_80_percent") model = SparseMistral.from_pretrained("thrunlab/Mistral_Sparse_pretraining_80_percent")