mteb-pt/average_pt_nilc_wang2vec_skip_s1000
This is an adaptation of pre-trained Portuguese Wang2Vec Word Embeddings to a sentence-transformers model.
The original pre-trained word embeddings can be found at: http://nilc.icmc.usp.br/nilc/index.php/repositorio-de-word-embeddings-do-nilc.
This model maps sentences & paragraphs to a 1000 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('mteb-pt/average_pt_nilc_wang2vec_skip_s1000')
embeddings = model.encode(sentences)
print(embeddings)
Evaluation Results
For an automated evaluation of this model, see the Portuguese MTEB Leaderboard: mteb-pt/leaderboard
Full Model Architecture
SentenceTransformer(
(0): WordEmbeddings(
(emb_layer): Embedding(929607, 1000)
)
(1): Pooling({'word_embedding_dimension': 1000, '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
@inproceedings{hartmann2017portuguese,
title = {Portuguese Word Embeddings: Evaluating on Word Analogies and Natural Language Tasks},
author = {Hartmann, Nathan S and
Fonseca, Erick R and
Shulby, Christopher D and
Treviso, Marcos V and
Rodrigues, J{'{e}}ssica S and
Alu{'{\i}}sio, Sandra Maria},
year = {2017},
publisher = {SBC},
booktitle = {Brazilian Symposium in Information and Human Language Technology - STIL},
url = {https://sol.sbc.org.br/index.php/stil/article/view/4008}
}
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.