bge-m3_en_ru

Uso (Sentence-Transformers)

pip install -U sentence-transformers

from sentence_transformers import SentenceTransformer
sentences = ["This is an example sentence", "Each sentence is converted"]

model = SentenceTransformer('TatonkaHF/bge-m3_en_ru')
embeddings = model.encode(sentences)
print(embeddings)

Uso (HuggingFace Transformers)

from transformers import AutoTokenizer, AutoModel
import torch

# Media de Pooling - Tomar en cuenta la máscara de atención para un promedio correcto
def mean_pooling(model_output, attention_mask):
    token_embeddings = model_output[0]  # El primer elemento de model_output contiene todas las embeddings de los tokens
    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)

# Frases para las cuales queremos embeddings de la oración
sentences = ['This is an example sentence', 'Each sentence is converted']

# Cargar el modelo desde HuggingFace Hub
tokenizer = AutoTokenizer.from_pretrained('TatonkaHF/bge-m3_en_ru')
model = AutoModel.from_pretrained('TatonkaHF/bge-m3_en_ru')

# Tokenizar las frases
encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')

# Calcular las embeddings de los tokens
with torch.no_grad():
    model_output = model(**encoded_input)

# Realizar pooling. En este caso, media de pooling.
sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])

print("Embeddings de las oraciones:")
print(sentence_embeddings)