yahyaabd / allstats-ir-mpnet-base-v1

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-statictable-query-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-statictable-query-title-pairs

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/allstats-ir-mpnet-base-v1")
# Run inference
sentences = [
    'Laporan keuangan perusahaan asuransi wajib & BPJS akhir 2015',
    'Ringkasan Neraca Arus Dana, Triwulan I, 2013*), (Miliar Rupiah)',
    'Rata-rata Konsumsi dan Pengeluaran Perkapita Seminggu Menurut Komoditi Makanan dan Golongan Pengeluaran per Kapita Seminggu di Provinsi Jawa Timur, 2018-2023',
]
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: allstats-semantic-base-v1-eval and allstat-semantic-base-v1-test
• Evaluated with EmbeddingSimilarityEvaluator

Training Details

Training Dataset

bps-statictable-query-title-pairs

• Dataset: bps-statictable-query-title-pairs at c7df38f
• Size: 2,602 training samples
• Columns: query , doc , and label
• Approximate statistics based on the first 1000 samples:

  	query	doc	label
  type	string	string	int
  details	min: 5 tokens mean: 18.35 tokens max: 35 tokens	min: 4 tokens mean: 25.83 tokens max: 58 tokens	0: ~66.50% 1: ~33.50%

• Samples:

  query	doc	label
  Pertumbuhan populasi provinsi di Indonesia 1971-2024	Kecepatan Angin dan Kelembaban di Stasiun Pengamatan BMKG, 2000-2010	0
  Perbandingan upah nominal dan riil pekerja pertanian di Indonesia (tahun dasar 2012), periode 2017.	Upah Nominal dan Riil Buruh Tani di Indonesia (Rupiah), 2009-2019 (2012=100)	1
  Laporan singkat cash flow statement Q4/2005	Nilai Produksi dan Biaya Produksi per Hektar Usaha Tanaman Bawang Merah dan Cabai Merah, 2014	0

• Loss: ContrastiveLoss with these parameters:

  {
      "distance_metric": "SiameseDistanceMetric.COSINE_DISTANCE",
      "margin": 0.5,
      "size_average": true
  }
  

Evaluation Dataset

bps-statictable-query-title-pairs

• Dataset: bps-statictable-query-title-pairs at c7df38f
• Size: 558 evaluation samples
• Columns: query , doc , and label
• Approximate statistics based on the first 558 samples:

  	query	doc	label
  type	string	string	int
  details	min: 4 tokens mean: 18.45 tokens max: 35 tokens	min: 4 tokens mean: 26.04 tokens max: 58 tokens	0: ~70.97% 1: ~29.03%

• Samples:

  query	doc	label
  Data pengeluaran makanan rata-rata warga Sulteng per minggu di tahun 2022, berdasarkan kelompok pendapatan	Sistem Neraca Sosial Ekonomi Indonesia Tahun 2022 (84 x 84)	0
  Konsumsi & belanja makanan per orang di NTB, beda kelompok pengeluaran, 2021	Rata-rata Konsumsi dan Pengeluaran Perkapita Seminggu Menurut Komoditi Makanan dan Golongan Pengeluaran per Kapita Seminggu di Provinsi Nusa Tenggara Barat, 2018-2023	1
  Bagaimana perbandingan PNS pria dan wanita di berbagai golongan tahun 2014?	Penduduk Berumur 15 Tahun Ke Atas Menurut Provinsi dan Jenis Kegiatan Selama Seminggu yang Lalu, 2008 - 2024	0

• Loss: ContrastiveLoss with these parameters:

  {
      "distance_metric": "SiameseDistanceMetric.COSINE_DISTANCE",
      "margin": 0.5,
      "size_average": true
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy : steps
• per_device_train_batch_size : 32
• per_device_eval_batch_size : 32
• num_train_epochs : 4
• warmup_ratio : 0.1
• fp16 : True
• load_best_model_at_end : True
• eval_on_start : True

All Hyperparameters

Training Logs

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.4.0
• Transformers: 4.48.1
• PyTorch: 2.5.1+cu124
• Accelerate: 1.3.0
• 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",
}

ContrastiveLoss

@inproceedings{hadsell2006dimensionality,
    author={Hadsell, R. and Chopra, S. and LeCun, Y.},
    booktitle={2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06)},
    title={Dimensionality Reduction by Learning an Invariant Mapping},
    year={2006},
    volume={2},
    number={},
    pages={1735-1742},
    doi={10.1109/CVPR.2006.100}
}