AbderrahmanSkiredj1/Arabic_text_embedding_for_sts

SentenceTransformer

This is a sentence-transformers model trained on the AbderrahmanSkiredj1/arabic_quora_duplicates_stsb_alue_holyquran_aranli_900k_anchor_positive_negative 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
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 tokens
• Similarity Function: Cosine Similarity
• Training Dataset:
  • AbderrahmanSkiredj1/arabic_quora_duplicates_stsb_alue_holyquran_aranli_900k_anchor_positive_negative

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': 512, 'do_lower_case': False}) with Transformer model: BertModel 
  (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("AbderrahmanSkiredj1/Arabic_text_embedding_for_sts")
# Run inference
sentences = [
    'يتم إنتاج أمثلة جميلة من المينا، والسيراميك، والفخار في وفرة كبيرة، وغالبا ما تتبع موضوع سلتيكي.',
    'يتم إنتاج عدد كبير من العناصر ذات المواضيع السلتية.',
    'يتم إنتاج الفخار الصغير الذي له موضوع سلتيكي.',
]
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]

Training Details

Training Dataset

AbderrahmanSkiredj1/arabic_quora_duplicates_stsb_alue_holyquran_aranli_900k_anchor_positive_negative

• Dataset: AbderrahmanSkiredj1/arabic_quora_duplicates_stsb_alue_holyquran_aranli_900k_anchor_positive_negative
• Size: 853,827 training samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 4 tokens mean: 14.54 tokens max: 91 tokens	min: 4 tokens mean: 10.62 tokens max: 43 tokens	min: 4 tokens mean: 10.32 tokens max: 35 tokens

• Samples:

  anchor	positive	negative
  هل يمكنك أن تأكل نفس الشيء كل يوم وتحصل على كل التغذية التي تحتاجها؟	هل الأكل نفس الشيء كل يوم صحي؟	ما هي القوة الخارقة التي تتمنى أن تملكها؟
  ثلاثة لاعبي كرة قدم، رقم 16 يرمي الكرة، رقم 71 يمنع الخصم الآخر.	لاعبي كرة القدم يرمون ويمنعون بعضهم البعض	الفريق يأكل البيتزا في مطعم
  كيف تحسن مهاراتك في الكتابة؟	كيف أستمر في تحسين كتابتي؟	كيف يتم تحديد أرقام الضمان الاجتماعي؟

• Loss: MatryoshkaLoss with these parameters:

  {
      "loss": "MultipleNegativesRankingLoss",
      "matryoshka_dims": [
          768,
          512,
          256,
          128,
          64
      ],
      "matryoshka_weights": [
          1,
          1,
          1,
          1,
          1
      ],
      "n_dims_per_step": -1
  }
  

Evaluation Dataset

AbderrahmanSkiredj1/arabic_quora_duplicates_stsb_alue_holyquran_aranli_900k_anchor_positive_negative

• Dataset: AbderrahmanSkiredj1/arabic_quora_duplicates_stsb_alue_holyquran_aranli_900k_anchor_positive_negative
• Size: 11,584 evaluation samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 3 tokens mean: 16.03 tokens max: 88 tokens	min: 3 tokens mean: 11.72 tokens max: 221 tokens	min: 3 tokens mean: 10.59 tokens max: 42 tokens

• Samples:

  anchor	positive	negative
  ماذا سيحدث لو توقفت الأرض عن الدوران وتدور في نفس الوقت؟	ماذا سيحدث إذا توقفت الأرض عن الدوران؟	ما هو أفضل هاتف ذكي تحت 15000؟
  ثلاثة متفرجين بالغين وطفل واحد ينظرون إلى السماء بينما يقفون على الرصيف.	أربعة أشخاص ينظرون إلى السماء.	رجل وثلاثة أطفال يشاهدون بالونات الهيليوم تطفو أعلى في الهواء
  ماذا تفعل الدول لمنع الحرب؟	كيف يجب على الدول أن تمنع الحرب؟	كيف يمكنني كسب المال من بدء مدونة؟

• Loss: MatryoshkaLoss with these parameters:

  {
      "loss": "MultipleNegativesRankingLoss",
      "matryoshka_dims": [
          768,
          512,
          256,
          128,
          64
      ],
      "matryoshka_weights": [
          1,
          1,
          1,
          1,
          1
      ],
      "n_dims_per_step": -1
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 64
• per_device_eval_batch_size: 64
• learning_rate: 1e-06
• num_train_epochs: 10
• warmup_ratio: 0.1
• fp16: True
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• 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
• learning_rate: 1e-06
• weight_decay: 0
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1.0
• num_train_epochs: 10
• 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
• 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: 0
• dataloader_prefetch_factor: None
• past_index: -1
• disable_tqdm: False
• remove_unused_columns: True
• label_names: None
• load_best_model_at_end: False
• 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}
• 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: False
• hub_always_push: False
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• include_inputs_for_metrics: False
• 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_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss
0.0120	40	3.1459
0.0240	80	3.2058
0.0360	120	3.0837
0.0480	160	3.1024
0.0600	200	3.015
0.0719	240	3.1311
0.0839	280	3.1101
0.0959	320	3.1288
0.1079	360	3.045
0.1199	400	3.0488
0.1319	440	3.1001
0.1439	480	3.2334
0.1559	520	3.0581
0.1679	560	2.9821
0.1799	600	3.1733
0.1918	640	3.0658
0.2038	680	3.0721
0.2158	720	3.1647
0.2278	760	3.0326
0.2398	800	3.1014
0.2518	840	2.9365
0.2638	880	3.0642
0.2758	920	2.9864
0.2878	960	3.0939
0.2998	1000	3.0676
0.3118	1040	2.9717
0.3237	1080	2.9908
0.3357	1120	2.9506
0.3477	1160	2.907
0.3597	1200	3.0451
0.3717	1240	3.0002
0.3837	1280	2.8842
0.3957	1320	3.0697
0.4077	1360	2.8967
0.4197	1400	3.0008
0.4317	1440	3.0027
0.4436	1480	2.9229
0.4556	1520	2.9539
0.4676	1560	2.9415
0.4796	1600	2.9401
0.4916	1640	2.8498
0.5036	1680	2.9646
0.5156	1720	2.9231
0.5276	1760	2.942
0.5396	1800	2.8521
0.5516	1840	2.8362
0.5635	1880	2.8497
0.5755	1920	2.8867
0.5875	1960	2.9148
0.5995	2000	2.9343
0.6115	2040	2.8537
0.6235	2080	2.7989
0.6355	2120	2.8508
0.6475	2160	2.916
0.6595	2200	2.926
0.6715	2240	2.752
0.6835	2280	2.7792
0.6954	2320	2.8381
0.7074	2360	2.7455
0.7194	2400	2.8953
0.7314	2440	2.8179
0.7434	2480	2.8471
0.7554	2520	2.7538
0.7674	2560	2.8271
0.7794	2600	2.8401
0.7914	2640	2.7402
0.8034	2680	2.6439

Framework Versions

• Python: 3.10.14
• Sentence Transformers: 3.0.1
• Transformers: 4.39.3
• PyTorch: 2.2.2+cu121
• Accelerate: 0.29.1
• Datasets: 2.18.0
• Tokenizers: 0.15.2

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",
}

MatryoshkaLoss

@misc{kusupati2024matryoshka,
    title={Matryoshka Representation Learning}, 
    author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
    year={2024},
    eprint={2205.13147},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

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}
}