yahyaabd/paraphrase-multilingual-miniLM-L12-V2-mnrl-2

SentenceTransformer based on sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2

This is a sentence-transformers model finetuned from sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2 on the statictable-triplets-all dataset. It maps sentences & paragraphs to a 384-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-MiniLM-L12-v2
• Maximum Sequence Length: 128 tokens
• Output Dimensionality: 384 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • statictable-triplets-all

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: BertModel 
  (1): Pooling({'word_embedding_dimension': 384, '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/paraphrase-multilingual-miniLM-L12-V2-mnrl-2")
# Run inference
sentences = [
    'Perbandingan Indeks Harga Konsumen antar negara, data tahun 2011',
    'Indeks Harga Konsumen Beberapa Negara (2005=100), 2009-2012',
    'Rata-rata Konsumsi dan Pengeluaran Perkapita Seminggu Menurut Komoditi Makanan dan Golongan Pengeluaran per Kapita Seminggu di Provinsi Kepulauan Riau, 2018-2023',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

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

Evaluation

Metrics

Information Retrieval

• Dataset: bps-statictable-ir
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.9121
cosine_accuracy@5	0.9837
cosine_accuracy@10	0.9837
cosine_precision@1	0.9121
cosine_precision@5	0.2254
cosine_precision@10	0.1235
cosine_recall@1	0.7068
cosine_recall@5	0.7895
cosine_recall@10	0.7967
cosine_ndcg@1	0.9121
cosine_ndcg@5	0.825
cosine_ndcg@10	0.8137
cosine_mrr@1	0.9121
cosine_mrr@5	0.9421
cosine_mrr@10	0.9421
cosine_map@1	0.9121
cosine_map@5	0.7806
cosine_map@10	0.7663

Training Details

Training Dataset

statictable-triplets-all

• Dataset: statictable-triplets-all at 24979b4
• Size: 967,831 training samples
• Columns: query, pos, and neg
• Approximate statistics based on the first 1000 samples:

  	query	pos	neg
  type	string	string	string
  details	min: 4 tokens mean: 18.52 tokens max: 37 tokens	min: 4 tokens mean: 25.18 tokens max: 49 tokens	min: 4 tokens mean: 25.87 tokens max: 58 tokens

• Samples:

  query	pos	neg
  Pekerja 15+ berdasarkan status pekerjaan & pendidikan, 1997-2007	Penduduk Berumur 15 Tahun Ke Atas yang Bekerja Selama Seminggu yang Lalu Menurut Status Pekerjaan Utama dan Pendidikan Tertinggi yang Ditamatkan, 1997 - 2007	Rata-rata Pendapatan Bersih Berusaha Sendiri Menurut Provinsi dan Kelompok Umur, 2023
  Output kilang minyak dan gas Indonesia berdasarkan jenis produk, dalam barel, sekitar tahun 2008	Produksi Beberapa Hasil Kilang Minyak dan Gas Menurut Jenis Hasil Kilang (barel), 2000-2018	Rata-rata Pendapatan Bersih Pekerja Bebas Menurut Provinsi dan Lapangan Pekerjaan Utama, 2021
  Berapa rata-rata pendapatan (ribu rupiah) pekerja bebas di Indonesia tahun 2018 berdasarkan provinsi dan pendidikan terakhirnya?	Rata-rata Pendapatan Bersih Pekerja Bebas Menurut Provinsi dan Pendidikan Tertinggi yang Ditamatkan (ribu rupiah), 2018	Rata-rata Upah/Gaji Bersih Sebulan Buruh/Karyawan/Pegawai Menurut Provinsi dan Lapangan Pekerjaan Utama di 17 Sektor (rupiah), 2018

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim"
  }
  

Evaluation Dataset

statictable-triplets-all

• Dataset: statictable-triplets-all at 24979b4
• Size: 967,831 evaluation samples
• Columns: query, pos, and neg
• Approximate statistics based on the first 1000 samples:

  	query	pos	neg
  type	string	string	string
  details	min: 5 tokens mean: 18.46 tokens max: 37 tokens	min: 4 tokens mean: 25.44 tokens max: 58 tokens	min: 4 tokens mean: 25.54 tokens max: 50 tokens

• Samples:

  query	pos	neg
  Data pendapatan rata-rata pengusaha di tiap provinsi berdasarkan tingkat pendidikan (tahun 2022)	Rata-rata Pendapatan bersih Berusaha Sendiri menurut Provinsi dan Pendidikan yang Ditamatkan, 2022	Kecepatan Angin dan Kelembaban di Stasiun Pengamatan BMKG, 2011-2015
  Berapa harga rata-rata valuta asing di berbagai provinsi Indonesia pada tahun 2017?	Rata-Rata Harga Valuta Asing Terpilih menurut Provinsi 2017	Kumulatif Kasus AIDS, Kasus Meninggal, Rate Kumulatif, dan Jumlah Kasus Baru AIDS Menurut Provinsi di Indonesia, 2008-2012
  Ikhtisar arus kas triwulan 1, 2004 (miliar)	Ringkasan Neraca Arus Dana Triwulan I 2004 (Miliar Rupiah)	Ekspor Bijih Tembaga Menurut Negara Tujuan Utama, 2012-2023

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• num_train_epochs: 1
• warmup_ratio: 0.1
• save_on_each_node: True
• fp16: True
• dataloader_num_workers: 2
• load_best_model_at_end: True
• eval_on_start: True
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• per_gpu_train_batch_size: None
• per_gpu_eval_batch_size: None
• gradient_accumulation_steps: 1
• eval_accumulation_steps: None
• torch_empty_cache_steps: None
• learning_rate: 5e-05
• weight_decay: 0.0
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1.0
• num_train_epochs: 1
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.1
• warmup_steps: 0
• log_level: passive
• log_level_replica: warning
• log_on_each_node: True
• logging_nan_inf_filter: True
• save_safetensors: True
• save_on_each_node: True
• save_only_model: False
• restore_callback_states_from_checkpoint: False
• no_cuda: False
• use_cpu: False
• use_mps_device: False
• seed: 42
• data_seed: None
• jit_mode_eval: False
• use_ipex: False
• bf16: False
• fp16: True
• fp16_opt_level: O1
• half_precision_backend: auto
• bf16_full_eval: False
• fp16_full_eval: False
• tf32: None
• local_rank: 0
• ddp_backend: None
• tpu_num_cores: None
• tpu_metrics_debug: False
• debug: []
• dataloader_drop_last: False
• dataloader_num_workers: 2
• dataloader_prefetch_factor: None
• past_index: -1
• disable_tqdm: False
• remove_unused_columns: True
• label_names: None
• load_best_model_at_end: True
• ignore_data_skip: False
• fsdp: []
• fsdp_min_num_params: 0
• fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
• fsdp_transformer_layer_cls_to_wrap: None
• accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
• deepspeed: None
• label_smoothing_factor: 0.0
• optim: adamw_torch
• optim_args: None
• adafactor: False
• group_by_length: False
• length_column_name: length
• ddp_find_unused_parameters: None
• ddp_bucket_cap_mb: None
• ddp_broadcast_buffers: False
• dataloader_pin_memory: True
• dataloader_persistent_workers: False
• skip_memory_metrics: True
• use_legacy_prediction_loop: False
• push_to_hub: False
• resume_from_checkpoint: None
• hub_model_id: None
• hub_strategy: every_save
• hub_private_repo: None
• hub_always_push: False
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• include_inputs_for_metrics: False
• include_for_metrics: []
• eval_do_concat_batches: True
• fp16_backend: auto
• push_to_hub_model_id: None
• push_to_hub_organization: None
• mp_parameters:
• auto_find_batch_size: False
• full_determinism: False
• torchdynamo: None
• ray_scope: last
• ddp_timeout: 1800
• torch_compile: False
• torch_compile_backend: None
• torch_compile_mode: None
• dispatch_batches: None
• split_batches: None
• include_tokens_per_second: False
• include_num_input_tokens_seen: False
• neftune_noise_alpha: None
• optim_target_modules: None
• batch_eval_metrics: False
• eval_on_start: True
• use_liger_kernel: False
• eval_use_gather_object: False
• average_tokens_across_devices: False
• prompts: None
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	Validation Loss	bps-statictable-ir_cosine_ndcg@10
0	0	-	1.1582	0.4643
0.0209	250	0.4825	0.0930	0.7864
0.0417	500	0.0746	0.0339	0.8081
0.0626	750	0.0337	0.0239	0.8064
0.0835	1000	0.0226	0.0149	0.8140
0.1044	1250	0.0174	0.0168	0.8267
0.1252	1500	0.0152	0.0185	0.8015
0.1461	1750	0.0127	0.0142	0.8269
0.1670	2000	0.0091	0.0104	0.8394
0.1879	2250	0.0089	0.0126	0.8255
0.2087	2500	0.0055	0.0133	0.8299
0.2296	2750	0.008	0.0086	0.8293
0.2505	3000	0.0087	0.0090	0.8317
0.2714	3250	0.0056	0.0061	0.8221
0.2922	3500	0.0057	0.0064	0.8220
0.3131	3750	0.0074	0.0056	0.8174
0.3340	4000	0.0033	0.0048	0.8263
0.3548	4250	0.0028	0.0054	0.8263
0.3757	4500	0.0036	0.0052	0.8055
0.3966	4750	0.0037	0.0037	0.8017
0.4175	5000	0.0045	0.0039	0.8242
0.4383	5250	0.0018	0.0035	0.8206
0.4592	5500	0.0013	0.0051	0.8202
0.4801	5750	0.0033	0.0033	0.8239
0.5010	6000	0.0017	0.0021	0.8210
0.5218	6250	0.0055	0.0119	0.8222
0.5427	6500	0.0019	0.0016	0.8070
0.5636	6750	0.001	0.0015	0.8128
0.5845	7000	0.0022	0.0013	0.8236
0.6053	7250	0.001	0.0005	0.8201
0.6262	7500	0.0005	0.0004	0.8226
0.6471	7750	0.001	0.0010	0.8251
0.6679	8000	0.0009	0.0011	0.8182
0.6888	8250	0.0015	0.0008	0.8185
0.7097	8500	0.0006	0.0005	0.8153
0.7306	8750	0.0012	0.0012	0.8133
0.7514	9000	0.0016	0.0007	0.8134
0.7723	9250	0.0011	0.0001	0.8152
0.7932	9500	0.0006	0.0002	0.8149
0.8141	9750	0.0011	0.0002	0.8141
0.8349	10000	0.0006	0.0002	0.8153
0.8558	10250	0.0002	0.0002	0.8174
0.8767	10500	0.0009	0.0001	0.8117
0.8976	10750	0.0006	0.0003	0.8117
0.9184	11000	0.0006	0.0003	0.8122
0.9393	11250	0.0008	0.0003	0.8161
0.9602	11500	0.0005	0.0003	0.8151
0.9810	11750	0.0004	0.0003	0.8137

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.11.11
• Sentence Transformers: 3.4.1
• Transformers: 4.48.3
• PyTorch: 2.5.1+cu124
• Accelerate: 1.3.0
• Datasets: 3.4.1
• 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",
}

MultipleNegativesRankingLoss

@misc{henderson2017efficient,
    title={Efficient Natural Language Response Suggestion for Smart Reply},
    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
    year={2017},
    eprint={1705.00652},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}