PORTULAN/serafim-900m-portuguese-pt-sentence-encoder

Serafim 900m Portuguese (PT) Sentence Encoder

This is a sentence-transformers model: It maps sentences & paragraphs to a 1536 dimensional dense vector space and can be used for tasks like clustering or semantic search.

Usage (Sentence-Transformers)

Using this model becomes easy when you have sentence-transformers installed:

pip install -U sentence-transformers

Then you can use the model like this:

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

model = SentenceTransformer('PORTULAN/serafim-900m-portuguese-pt-sentence-encoder')
embeddings = model.encode(sentences)
print(embeddings)

Usage (HuggingFace Transformers)

Without sentence-transformers, you can use the model like this: First, you pass your input through the transformer model, then you have to apply the right pooling-operation on-top of the contextualized word embeddings.

from transformers import AutoTokenizer, AutoModel
import torch


#Mean Pooling - Take attention mask into account for correct averaging
def mean_pooling(model_output, attention_mask):
    token_embeddings = model_output[0] #First element of model_output contains all token embeddings
    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)


# Sentences we want sentence embeddings for
sentences = ['This is an example sentence', 'Each sentence is converted']

# Load model from HuggingFace Hub
tokenizer = AutoTokenizer.from_pretrained('PORTULAN/serafim-900m-portuguese-pt-sentence-encoder')
model = AutoModel.from_pretrained('PORTULAN/serafim-900m-portuguese-pt-sentence-encoder')

# Tokenize sentences
encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')

# Compute token embeddings
with torch.no_grad():
    model_output = model(**encoded_input)

# Perform pooling. In this case, mean pooling.
sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])

print("Sentence embeddings:")
print(sentence_embeddings)

Evaluation Results

For an automated evaluation of this model, see the Sentence Embeddings Benchmark: https://seb.sbert.net

Training

The model was trained with the parameters:

DataLoader:

torch.utils.data.dataloader.DataLoader of length 1183 with parameters:

{'batch_size': 16, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}

Loss:

sentence_transformers.losses.CoSENTLoss.CoSENTLoss with parameters:

{'scale': 20.0, 'similarity_fct': 'pairwise_cos_sim'}

Parameters of the fit()-Method:

{
    "epochs": 10,
    "evaluation_steps": 119,
    "evaluator": "sentence_transformers.evaluation.EmbeddingSimilarityEvaluator.EmbeddingSimilarityEvaluator",
    "max_grad_norm": 1,
    "optimizer_class": "<class 'torch.optim.adamw.AdamW'>",
    "optimizer_params": {
        "lr": 1e-06
    },
    "scheduler": "WarmupLinear",
    "steps_per_epoch": 1183,
    "warmup_steps": 1183,
    "weight_decay": 0.01
}

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 128, 'do_lower_case': False}) with Transformer model: DebertaV2Model 
  (1): Pooling({'word_embedding_dimension': 1536, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
)

Citing & Authors

The article has been presented at EPIA 2024 conference but the Springer proceedings are not available yet. In the meantime, if you use this model you may cite the arXiv preprint:

@misc{gomes2024opensentenceembeddingsportuguese,
      title={Open Sentence Embeddings for Portuguese with the Serafim PT* encoders family}, 
      author={Luís Gomes and António Branco and João Silva and João Rodrigues and Rodrigo Santos},
      year={2024},
      eprint={2407.19527},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2407.19527}, 
}