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 sentence-transformers/quora-duplicates 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 Sources

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("DashReza7/paraphrase-multilingual-MiniLM-L12-v2_QuoraDuplicateDetection_FINETUNED")
# Run inference
sentences = [
    'Why do complementary angles have to be adjacent?',
    'Can two adjacent angles be complementary?',
    'How can I get rid of my bad habits?',
]
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

Binary Classification

Metric Value
cosine_accuracy 0.8684
cosine_accuracy_threshold 0.7981
cosine_f1 0.8292
cosine_f1_threshold 0.7599
cosine_precision 0.7747
cosine_recall 0.8921
cosine_ap 0.8822
dot_accuracy 0.836
dot_accuracy_threshold 17.1121
dot_f1 0.7914
dot_f1_threshold 16.0833
dot_precision 0.7294
dot_recall 0.865
dot_ap 0.8439
manhattan_accuracy 0.8568
manhattan_accuracy_threshold 46.9431
manhattan_f1 0.8144
manhattan_f1_threshold 50.5148
manhattan_precision 0.7656
manhattan_recall 0.8698
manhattan_ap 0.8636
euclidean_accuracy 0.8569
euclidean_accuracy_threshold 3.0017
euclidean_f1 0.8143
euclidean_f1_threshold 3.2429
euclidean_precision 0.7652
euclidean_recall 0.8701
euclidean_ap 0.8638
max_accuracy 0.8684
max_accuracy_threshold 46.9431
max_f1 0.8292
max_f1_threshold 50.5148
max_precision 0.7747
max_recall 0.8921
max_ap 0.8822

Training Details

Training Dataset

sentence-transformers/quora-duplicates

  • Dataset: sentence-transformers/quora-duplicates at 451a485
  • Size: 323,432 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: 16.39 tokens
    • max: 80 tokens
    • min: 4 tokens
    • mean: 16.2 tokens
    • max: 71 tokens
    • 0: ~62.10%
    • 1: ~37.90%
  • Samples:
    sentence1 sentence2 label
    Which are the best compilers for C language (for Windows 10)? Which is the best open source C/C++ compiler for Windows? 0
    How much does YouTube pay per 1000 views in India? How much does youtube pay per 1000 views? 0
    What parts do I need to build my own PC? I want to build a new computer. What parts do I need? 1
  • Loss: OnlineContrastiveLoss

Evaluation Dataset

sentence-transformers/quora-duplicates

  • Dataset: sentence-transformers/quora-duplicates at 451a485
  • Size: 80,858 evaluation 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: 16.48 tokens
    • max: 79 tokens
    • min: 6 tokens
    • mean: 16.76 tokens
    • max: 101 tokens
    • 0: ~63.90%
    • 1: ~36.10%
  • Samples:
    sentence1 sentence2 label
    How many stories got busted on Quora while being anonymous? Can what I say on Quora anonymously be used against me legally? 0
    What are innovative mechanical component designs? What is the Innovation design? 0
    What is the best way to learn phrasal verbs? Why should I learn phrasal verbs? 1
  • Loss: OnlineContrastiveLoss

Training Hyperparameters

Non-Default Hyperparameters

  • eval_strategy: steps
  • per_device_train_batch_size: 256
  • per_device_eval_batch_size: 256
  • learning_rate: 2e-05
  • num_train_epochs: 1
  • warmup_ratio: 0.1
  • fp16: 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: 256
  • per_device_eval_batch_size: 256
  • per_gpu_train_batch_size: None
  • per_gpu_eval_batch_size: None
  • gradient_accumulation_steps: 1
  • eval_accumulation_steps: None
  • learning_rate: 2e-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: 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: False
  • hub_always_push: False
  • gradient_checkpointing: False
  • gradient_checkpointing_kwargs: None
  • include_inputs_for_metrics: False
  • 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
  • batch_sampler: batch_sampler
  • multi_dataset_batch_sampler: proportional

Training Logs

Epoch Step Training Loss loss max_ap
0.0791 100 - 8.0607 0.8164
0.1582 200 - 7.3012 0.8445
0.2373 300 - 6.9626 0.8582
0.3165 400 - 6.7901 0.8639
0.3956 500 7.5229 6.6498 0.8694
0.4747 600 - 6.5315 0.8736
0.5538 700 - 6.4686 0.8766
0.6329 800 - 6.4027 0.8787
0.7120 900 - 6.3108 0.8797
0.7911 1000 6.4636 6.2862 0.8812
0.8703 1100 - 6.2449 0.8818
0.9494 1200 - 6.2344 0.8822

Framework Versions

  • Python: 3.10.12
  • Sentence Transformers: 3.0.1
  • Transformers: 4.42.4
  • PyTorch: 2.3.1+cu121
  • Accelerate: 0.32.1
  • Datasets: 2.21.0
  • Tokenizers: 0.19.1

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",
}
Downloads last month
17
Safetensors
Model size
118M params
Tensor type
F32
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for DashReza7/paraphrase-multilingual-MiniLM-L12-v2_QuoraDuplicateDetection_FINETUNED

Dataset used to train DashReza7/paraphrase-multilingual-MiniLM-L12-v2_QuoraDuplicateDetection_FINETUNED

Evaluation results