melino2000/product-torob-matching

SentenceTransformer based on sentence-transformers/quora-distilbert-multilingual

This is a sentence-transformers model finetuned from sentence-transformers/quora-distilbert-multilingual. 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/quora-distilbert-multilingual
• Maximum Sequence Length: 128 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity

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: DistilBertModel 
  (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("melino2000/product-torob-matching")
# Run inference
sentences = [
    'رایزر گرافیک مدل 009s plus هشت خازنه',
    'رایزر گرافیک تبدیل PCI EXPRESS X1 به X16 مدل 009S',
    'شامپو کودک حاوی عصاره اسطوخودوس فیروز200 میل',
]
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

Binary Classification

• Dataset: product-matching-binary
• Evaluated with BinaryClassificationEvaluator

Metric	Value
cosine_accuracy	0.9908
cosine_accuracy_threshold	0.7373
cosine_f1	0.9908
cosine_f1_threshold	0.7295
cosine_precision	0.99
cosine_recall	0.9915
cosine_ap	0.9989
cosine_mcc	0.9815

Training Details

Training Dataset

Unnamed Dataset

• Size: 32,000 training samples
• Columns: sentence1, sentence2, and label
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2	label
  type	string	string	int
  details	min: 6 tokens mean: 19.0 tokens max: 68 tokens	min: 5 tokens mean: 19.06 tokens max: 56 tokens	0: ~49.40% 1: ~50.60%

• Samples:

  sentence1	sentence2	label
  پرینتر چندکاره لیزری HP LaserJet Pro M130a	پرینتر لیزری سه کاره اچ پی HP M130a	1
  قرص روکشدار مولتی دیلی دکتر گیل 60 عددی داروسازی رازان فارمدیان	قرص مولتی دیلی دکتر گیل	1
  خمیردندان کلگیت 3 کاره Triple Action	100 میل	خمیر دندان کولگیت مدل 3 کاره حجم 100 میل

• Loss: OnlineContrastiveLoss

Evaluation Dataset

Unnamed Dataset

• Size: 8,000 evaluation samples
• Columns: sentence1, sentence2, and label
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2	label
  type	string	string	int
  details	min: 5 tokens mean: 19.24 tokens max: 69 tokens	min: 2 tokens mean: 18.76 tokens max: 56 tokens	0: ~50.00% 1: ~50.00%

• Samples:

  sentence1	sentence2	label
  مایکرو فر 36 لیتری ناسا الکتریک مدل NS-2024	سرویس کاور روتختی تک نفره ایکیا مدل Ikea BRUNKRISSLA 404.907.23	0
  کنسول بازی نینتندو سوییچ سفید - Nintendo Switch OLED Model white	NINTENDO SWITCH OLED (Neon Red & Neon Blue)	1
  خمیر دندان کرست مدل Complete 7	قلمو سرگرد 2122 پارس آرت (32400_107700 تومان)	0

• Loss: OnlineContrastiveLoss

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 32
• warmup_ratio: 0.1
• fp16: 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: 32
• per_device_eval_batch_size: 32
• 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: 3
• 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: 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, '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: False
• 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	product-matching-binary_cosine_ap
-1	-1	-	-	0.9436
0.1	100	0.637	-	-
0.2	200	0.1303	-	-
0.25	250	-	0.0785	0.9961
0.3	300	0.1378	-	-
0.4	400	0.1191	-	-
0.5	500	0.0949	0.0723	0.9963
0.6	600	0.1016	-	-
0.7	700	0.0694	-	-
0.75	750	-	0.0464	0.9974
0.8	800	0.0619	-	-
0.9	900	0.0543	-	-
1.0	1000	0.0658	0.0394	0.9981
1.1	1100	0.0326	-	-
1.2	1200	0.0176	-	-
1.25	1250	-	0.0387	0.9980
1.3	1300	0.0237	-	-
1.4	1400	0.0219	-	-
1.5	1500	0.0115	0.0259	0.9983
1.6	1600	0.0218	-	-
1.7	1700	0.0235	-	-
1.75	1750	-	0.0230	0.9988
1.8	1800	0.0319	-	-
1.9	1900	0.0127	-	-
2.0	2000	0.015	0.0285	0.9987
2.099	2100	0.0121	-	-
2.199	2200	0.0091	-	-
2.249	2250	-	0.0217	0.9986
2.299	2300	0.0107	-	-
2.399	2400	0.009	-	-
2.499	2500	0.0043	0.0224	0.9989
2.599	2600	0.0028	-	-
2.699	2700	0.0026	-	-
2.749	2750	-	0.0248	0.9989
2.799	2800	0.0024	-	-
2.899	2900	0.0067	-	-
2.999	3000	0.0088	0.0225	0.9989
-1	-1	-	-	0.9989

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: 3.3.2
• 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",
}