Alibaba-NLP

/

gte-base-en-v1.5

like 34

Hi @keerthi1306logesh , you can use the eager attention in M3 chip machines. By default, the xformers are not needed.

Load the model in this way:

from transformers import AutoTokenizer,  AutoModel

tokenizer = AutoTokenizer.from_pretrained('YOUR_MODEL')
model = AutoModel.from_pretrained('YOUR_MODEL', attn_implementation='eager', trust_remote_code=True)