How to load with HF Transformers?
#17
by
jhflow
- opened
Hi, Thank you for your remarkable work!. I'm really impressed by the performance of this model.
For some reason, I want to load this model via Huggingface transformers (AutoModel.from_pretrinaed or something) not via FlagEmbdding.
Can I do so?
Yes, you can load it in the same way with bge-1.5: https://github.com/FlagOpen/FlagEmbedding/tree/master/FlagEmbedding/baai_general_embedding#using-huggingface-transformers
jhflow
changed discussion status to
closed
Thank you!
How can I get dense, colbert embeddings with transformers?
Given
from transformers import AutoModel, AutoTokenizer
from torch import Tensor
import torch
model_path = 'BAAI/bge-m3'
model = AutoModel.from_pretrained(model_path)
tokenizer = AutoTokenizer.from_pretrained(model_path)
test_sentence = ["this is a test sentence"]
batch_dict = tokenizer(test_sentence, return_tensors='pt', max_length=128, padding=True, truncation=True)
outputs = model(**batch_dict)
I get BaseModelOutputWithPoolingAndCrossAttentions with pooler_output and last_hidden_state keys. Is pooler_output the CLS embedding and last_hidden_state all the token embeddings?
Kindly clarify. Thank you.