yahyaabd / allstat-semantic-search-mpnet-base-v2-sts

SentenceTransformer based on sentence-transformers/paraphrase-multilingual-mpnet-base-v2

This is a sentence-transformers model finetuned from sentence-transformers/paraphrase-multilingual-mpnet-base-v2 on the bps-publication-title-pairs dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: sentence-transformers/paraphrase-multilingual-mpnet-base-v2
• Maximum Sequence Length: 128 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • bps-publication-title-pairs
• Language: id

Model Sources

• Documentation: Sentence Transformers Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Sentence Transformers on Hugging Face

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 128, 'do_lower_case': False}) with Transformer model: XLMRobertaModel 
  (1): Pooling({'word_embedding_dimension': 768, '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})
)

Usage

Direct Usage (Sentence Transformers)

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("yahyaabd/allstat-semantic-search-mpnet-base-v2-sts")
# Run inference
sentences = [
    'Laporan keuangan pemerintah provinsi periode 2003-2006',
    'Statistik Keuangan Provinsi 2003-2006',
    'Statistik Perdagangan Luar Negeri Indonesia Ekspor Menurut Kode ISIC 2013-2014',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Evaluation

Metrics

Semantic Similarity

• Datasets: allstat-semantic-dev and allstat-semantic-test
• Evaluated with EmbeddingSimilarityEvaluator

Training Details

Training Dataset

bps-publication-title-pairs

• Dataset: bps-publication-title-pairs at 833f738
• Size: 42,138 training samples
• Columns: query , doc_title , and score
• Approximate statistics based on the first 1000 samples:

  	query	doc_title	score
  type	string	string	float
  details	min: 5 tokens mean: 10.71 tokens max: 60 tokens	min: 5 tokens mean: 12.58 tokens max: 52 tokens	min: 0.0 mean: 0.53 max: 1.0

• Samples:

  query	doc_title	score
  Hasil riset mobilitas Jabodetabek tahun 2023	Statistik Komuter Jabodetabek Hasil Survei Komuter Jabodetabek 2023	0.85
  Indeks harga konsumen di Indonesia tahun 2017 (82 kota)	Harga Konsumen Beberapa Barang dan Jasa Kelompok Sandang di 82 Kota di Indonesia 2017	0.15
  Laporan sektor bangunan Indonesia Q4 2009	Indikator Konstruksi Triwulan IV Tahun 2009	0.91

• Loss: CosineSimilarityLoss with these parameters:

  {
      "loss_fct": "torch.nn.modules.loss.MSELoss"
  }
  

Evaluation Dataset

bps-publication-title-pairs

• Dataset: bps-publication-title-pairs at 833f738
• Size: 2,634 evaluation samples
• Columns: query , doc_title , and score
• Approximate statistics based on the first 1000 samples:

  	query	doc_title	score
  type	string	string	float
  details	min: 6 tokens mean: 10.71 tokens max: 28 tokens	min: 5 tokens mean: 12.57 tokens max: 39 tokens	min: 0.0 mean: 0.55 max: 1.0

• Samples:

  query	doc_title	score
  Statistik tebu Indonesia tahun 2018	Direktori Perusahaan Perkebunan Karet Indonesia 2018	0.1
  Data industri makanan dan minuman 2017	Statistik Upah Buruh Tani di Perdesaan 2018	0.2
  Biaya hidup di Gorontalo tahun 2018	Survei Biaya Hidup (SBH) 2018 Gorontalo	0.9

• Loss: CosineSimilarityLoss with these parameters:

  {
      "loss_fct": "torch.nn.modules.loss.MSELoss"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy : steps
• per_device_train_batch_size : 16
• per_device_eval_batch_size : 16
• num_train_epochs : 5
• warmup_ratio : 0.1
• fp16 : True

All Hyperparameters

Training Logs

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.3.1
• Transformers: 4.47.1
• PyTorch: 2.2.2+cu121
• Accelerate: 1.2.1
• Datasets: 3.2.0
• Tokenizers: 0.21.0

Citation

BibTeX

Sentence Transformers

@inproceedings{reimers-2019-sentence-bert,
    title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/1908.10084",
}