yahyaabd/allstats-v1-1

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 allstats-search-pairs-dataset 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:
  • allstats-search-pairs-dataset

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-v1-1")
# Run inference
sentences = [
    'Biaya hidup kelompok perumahan Indonesia 2017',
    'Statistik Upah 2013',
    'Survei Biaya Hidup (SBH) 2018 Bulukumba, Watampone, Makassar, Pare-Pare, dan Palopo',
]
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-mpnet-eval and allstats-semantic-mpnet-test
• Evaluated with EmbeddingSimilarityEvaluator

Metric	allstats-semantic-mpnet-eval	allstats-semantic-mpnet-test
pearson_cosine	0.9833	0.9833
spearman_cosine	0.8515	0.8521

Training Details

Training Dataset

allstats-search-pairs-dataset

• Dataset: allstats-search-pairs-dataset at 6712cb1
• Size: 79,621 training samples
• Columns: query, doc, and label
• Approximate statistics based on the first 1000 samples:

  	query	doc	label
  type	string	string	float
  details	min: 5 tokens mean: 10.78 tokens max: 39 tokens	min: 5 tokens mean: 13.73 tokens max: 58 tokens	min: 0.0 mean: 0.44 max: 0.99

• Samples:

  query	doc	label
  Produksi jagung di Indonesia tahun 2009	Indeks Unit Value Ekspor Menurut Kode SITC Bulan Februari 2024	0.1
  Data produksi industri manufaktur 2021	Perkembangan Indeks Produksi Industri Manufaktur 2021	0.96
  direktori perusahaan industri penggilingan padi tahun 2012 provinsi sulawesi utara dan gorontalo	Neraca Pemerintahan Umum Indonesia 2007-2012	0.03

• Loss: CosineSimilarityLoss with these parameters:

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

Evaluation Dataset

allstats-search-pairs-dataset

• Dataset: allstats-search-pairs-dataset at 6712cb1
• Size: 9,952 evaluation samples
• Columns: query, doc, and label
• Approximate statistics based on the first 1000 samples:

  	query	doc	label
  type	string	string	float
  details	min: 5 tokens mean: 10.75 tokens max: 40 tokens	min: 4 tokens mean: 14.09 tokens max: 49 tokens	min: 0.01 mean: 0.48 max: 0.99

• Samples:

  query	doc	label
  Daftar perusahaan industri pengolahan skala kecil 2006	Statistik Migrasi Nusa Tenggara Barat Hasil SP 2010	0.05
  Populasi Indonesia per provinsi 2000-2010	Indikator Ekonomi Desember 2023	0.08
  Data harga barang desa non-pangan tahun 2022	Statistik Kunjungan Tamu Asing 2004	0.1

• 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: 64
• per_device_eval_batch_size: 64
• num_train_epochs: 12
• warmup_ratio: 0.1
• fp16: True
• dataloader_num_workers: 4
• load_best_model_at_end: True
• label_smoothing_factor: 0.01
• eval_on_start: True

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 64
• per_device_eval_batch_size: 64
• 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: 12
• 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: False
• 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: 4
• 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: 0.01
• 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: batch_sampler
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	Validation Loss	allstats-semantic-mpnet-eval_spearman_cosine	allstats-semantic-mpnet-test_spearman_cosine
0	0	-	0.0958	0.6404	-
0.2008	250	0.0464	0.0246	0.7693	-
0.4016	500	0.0218	0.0179	0.7720	-
0.6024	750	0.0172	0.0153	0.7790	-
0.8032	1000	0.0156	0.0136	0.7809	-
1.0040	1250	0.0137	0.0139	0.7769	-
1.2048	1500	0.0112	0.0120	0.7825	-
1.4056	1750	0.0104	0.0112	0.7869	-
1.6064	2000	0.01	0.0103	0.7893	-
1.8072	2250	0.009	0.0097	0.7944	-
2.0080	2500	0.0088	0.0097	0.7947	-
2.2088	2750	0.0064	0.0086	0.7971	-
2.4096	3000	0.006	0.0085	0.7991	-
2.6104	3250	0.006	0.0084	0.7995	-
2.8112	3500	0.006	0.0081	0.8047	-
3.0120	3750	0.0058	0.0082	0.8055	-
3.2129	4000	0.0041	0.0077	0.8096	-
3.4137	4250	0.0042	0.0078	0.8092	-
3.6145	4500	0.004	0.0074	0.8107	-
3.8153	4750	0.0043	0.0073	0.8132	-
4.0161	5000	0.0044	0.0076	0.8090	-
4.2169	5250	0.0032	0.0071	0.8173	-
4.4177	5500	0.0031	0.0068	0.8218	-
4.6185	5750	0.0031	0.0067	0.8200	-
4.8193	6000	0.0032	0.0065	0.8233	-
5.0201	6250	0.0029	0.0067	0.8227	-
5.2209	6500	0.0024	0.0064	0.8249	-
5.4217	6750	0.0023	0.0066	0.8298	-
5.6225	7000	0.0025	0.0063	0.8271	-
5.8233	7250	0.0024	0.0064	0.8299	-
6.0241	7500	0.0023	0.0064	0.8312	-
6.2249	7750	0.0017	0.0061	0.8319	-
6.4257	8000	0.0017	0.0059	0.8330	-
6.6265	8250	0.0019	0.0064	0.8309	-
6.8273	8500	0.002	0.0061	0.8332	-
7.0281	8750	0.0018	0.0061	0.8360	-
7.2289	9000	0.0014	0.0060	0.8387	-
7.4297	9250	0.0014	0.0059	0.8396	-
7.6305	9500	0.0014	0.0059	0.8402	-
7.8313	9750	0.0014	0.0059	0.8388	-
8.0321	10000	0.0014	0.0058	0.8411	-
8.2329	10250	0.0011	0.0059	0.8420	-
8.4337	10500	0.0011	0.0057	0.8431	-
8.6345	10750	0.0011	0.0057	0.8418	-
8.8353	11000	0.0011	0.0057	0.8440	-
9.0361	11250	0.0011	0.0057	0.8449	-
9.2369	11500	0.0008	0.0056	0.8451	-
9.4378	11750	0.0009	0.0057	0.8456	-
9.6386	12000	0.0009	0.0056	0.8469	-
9.8394	12250	0.0009	0.0056	0.8470	-
10.0402	12500	0.0009	0.0056	0.8475	-
10.2410	12750	0.0007	0.0056	0.8489	-
10.4418	13000	0.0007	0.0056	0.8495	-
10.6426	13250	0.0007	0.0056	0.8501	-
10.8434	13500	0.0007	0.0056	0.8497	-
11.0442	13750	0.0006	0.0056	0.8500	-
11.245	14000	0.0006	0.0055	0.8506	-
11.4458	14250	0.0006	0.0055	0.8507	-
11.6466	14500	0.0006	0.0055	0.8512	-
11.8474	14750	0.0006	0.0055	0.8515	-
12.0	14940	-	-	-	0.8521

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.3.1
• Transformers: 4.47.0
• PyTorch: 2.5.1+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",
}