tomaarsen/stsb-distilbert-base-ocl

SentenceTransformer based on sentence-transformers/stsb-distilbert-base

This is a sentence-transformers model finetuned from sentence-transformers/stsb-distilbert-base on the sentence-transformers/quora-duplicates 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/stsb-distilbert-base
• Maximum Sequence Length: 128 tokens
• Output Dimensionality: 768 tokens
• Similarity Function: Cosine Similarity
• Training Dataset:
  • sentence-transformers/quora-duplicates
• Language: en

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("tomaarsen/stsb-distilbert-base-ocl")
# Run inference
sentences = [
    'Is stretching bad?',
    'Is stretching good for you?',
    'If i=0; what will i=i++ do to i?',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

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

Evaluation

Metrics

Binary Classification

• Dataset: quora-duplicates
• Evaluated with BinaryClassificationEvaluator

Metric	Value
cosine_accuracy	0.86
cosine_accuracy_threshold	0.8104
cosine_f1	0.8251
cosine_f1_threshold	0.7248
cosine_precision	0.7347
cosine_recall	0.9407
cosine_ap	0.8872
dot_accuracy	0.828
dot_accuracy_threshold	157.3549
dot_f1	0.7899
dot_f1_threshold	145.7113
dot_precision	0.7155
dot_recall	0.8814
dot_ap	0.8369
manhattan_accuracy	0.868
manhattan_accuracy_threshold	208.0035
manhattan_f1	0.8308
manhattan_f1_threshold	208.0035
manhattan_precision	0.7922
manhattan_recall	0.8733
manhattan_ap	0.8868
euclidean_accuracy	0.867
euclidean_accuracy_threshold	9.2694
euclidean_f1	0.8301
euclidean_f1_threshold	9.5257
euclidean_precision	0.7888
euclidean_recall	0.876
euclidean_ap	0.8884
max_accuracy	0.868
max_accuracy_threshold	208.0035
max_f1	0.8308
max_f1_threshold	208.0035
max_precision	0.7922
max_recall	0.9407
max_ap	0.8884

Paraphrase Mining

• Dataset: quora-duplicates-dev
• Evaluated with ParaphraseMiningEvaluator

Metric	Value
average_precision	0.5344
f1	0.5448
precision	0.5311
recall	0.5592
threshold	0.8626

Information Retrieval

• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.928
cosine_accuracy@3	0.9712
cosine_accuracy@5	0.9782
cosine_accuracy@10	0.9874
cosine_precision@1	0.928
cosine_precision@3	0.4151
cosine_precision@5	0.2666
cosine_precision@10	0.1417
cosine_recall@1	0.7994
cosine_recall@3	0.9342
cosine_recall@5	0.9561
cosine_recall@10	0.9766
cosine_ndcg@10	0.9516
cosine_mrr@10	0.9509
cosine_map@100	0.939
dot_accuracy@1	0.8926
dot_accuracy@3	0.9518
dot_accuracy@5	0.9658
dot_accuracy@10	0.9768
dot_precision@1	0.8926
dot_precision@3	0.4027
dot_precision@5	0.2608
dot_precision@10	0.1388
dot_recall@1	0.768
dot_recall@3	0.9106
dot_recall@5	0.9402
dot_recall@10	0.9623
dot_ndcg@10	0.9264
dot_mrr@10	0.9243
dot_map@100	0.9094

Training Details

Training Dataset

sentence-transformers/quora-duplicates

• Dataset: sentence-transformers/quora-duplicates at 451a485
• Size: 100,000 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: 15.5 tokens max: 45 tokens	min: 6 tokens mean: 15.46 tokens max: 78 tokens	0: ~64.10% 1: ~35.90%

• Samples:

  sentence1	sentence2	label
  What are the best ecommerce blogs to do guest posts on about SEO to gain new clients?	Interested in being a guest blogger for an ecommerce marketing blog?	0
  How do I learn Informatica online training?	What is Informatica online training?	0
  What effects does marijuana use have on the flu?	What effects does Marijuana use have on the common cold?	0

• Loss: OnlineContrastiveLoss

Evaluation Dataset

sentence-transformers/quora-duplicates

• Dataset: sentence-transformers/quora-duplicates at 451a485
• Size: 1,000 evaluation samples
• Columns: sentence1, sentence2, and label
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2	label
  type	string	string	int
  details	min: 6 tokens mean: 15.82 tokens max: 46 tokens	min: 6 tokens mean: 15.91 tokens max: 72 tokens	0: ~62.90% 1: ~37.10%

• Samples:

  sentence1	sentence2	label
  How should I prepare for JEE Mains 2017?	How do I prepare for the JEE 2016?	0
  What is the gate exam?	What is the GATE exam in engineering?	0
  Where do IRS officers get posted?	Does IRS Officers get posted abroad?	0

• Loss: OnlineContrastiveLoss

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 64
• per_device_eval_batch_size: 64
• num_train_epochs: 1
• 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: False
• 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: 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: 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, '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: None
• 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_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	loss	cosine_map@100	quora-duplicates-dev_average_precision	quora-duplicates_max_ap
0	0	-	-	0.9235	0.4200	0.7276
0.0640	100	2.5123	-	-	-	-
0.1280	200	2.0534	-	-	-	-
0.1599	250	-	1.7914	0.9127	0.4082	0.8301
0.1919	300	1.9505	-	-	-	-
0.2559	400	1.9836	-	-	-	-
0.3199	500	1.8462	1.5923	0.9190	0.4445	0.8688
0.3839	600	1.7734	-	-	-	-
0.4479	700	1.7918	-	-	-	-
0.4798	750	-	1.5461	0.9291	0.4943	0.8707
0.5118	800	1.6157	-	-	-	-
0.5758	900	1.7244	-	-	-	-
0.6398	1000	1.7322	1.5294	0.9309	0.5048	0.8808
0.7038	1100	1.6825	-	-	-	-
0.7678	1200	1.6823	-	-	-	-
0.7997	1250	-	1.4812	0.9351	0.5126	0.8865
0.8317	1300	1.5707	-	-	-	-
0.8957	1400	1.6145	-	-	-	-
0.9597	1500	1.5795	1.4705	0.9390	0.5344	0.8884

Environmental Impact

Carbon emissions were measured using CodeCarbon.

• Energy Consumed: 0.040 kWh
• Carbon Emitted: 0.016 kg of CO2
• Hours Used: 0.202 hours

Training Hardware

• On Cloud: No
• GPU Model: 1 x NVIDIA GeForce RTX 3090
• CPU Model: 13th Gen Intel(R) Core(TM) i7-13700K
• RAM Size: 31.78 GB

Framework Versions

• Python: 3.11.6
• Sentence Transformers: 3.0.0.dev0
• Transformers: 4.41.0.dev0
• PyTorch: 2.3.0+cu121
• Accelerate: 0.26.1
• Datasets: 2.18.0
• 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",
}