README.md

metadata

tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:910013
  - loss:CosineSimilarityLoss
base_model: intfloat/multilingual-e5-small
widget:
  - source_sentence: business healing
    sentences:
      - modify ict system capacity
      - objetividade, inovadora,estudiosa,pesquisadora e organizada
      - business consulting
  - source_sentence: architecture acoustics
    sentences:
      - disicpline leader
      - 生产工艺开发及优化
      - data analysis
  - source_sentence: arbitru natatie
    sentences:
      - criação cinematográfica
      - quarterly distribution
      - улучшение путешествий клиентов с помощью дополненной реальности
  - source_sentence: configuración de software antivirus
    sentences:
      - protocol & coordination
      - laurea magistrale biologia
      - deploy anti-virus software
  - source_sentence: child maltreatment counselling
    sentences:
      - book covers, flyers, posters, banners
      - tool and die making
      - cmc
pipeline_tag: sentence-similarity
library_name: sentence-transformers
metrics:
  - pearson_cosine
  - spearman_cosine
model-index:
  - name: SentenceTransformer based on intfloat/multilingual-e5-small
    results:
      - task:
          type: semantic-similarity
          name: Semantic Similarity
        dataset:
          name: sts dev
          type: sts-dev
        metrics:
          - type: pearson_cosine
            value: 0.9579653395486292
            name: Pearson Cosine
          - type: spearman_cosine
            value: 0.8788941637037295
            name: Spearman Cosine
      - task:
          type: semantic-similarity
          name: Semantic Similarity
        dataset:
          name: sts test
          type: sts-test
        metrics:
          - type: pearson_cosine
            value: 0.9579215714676803
            name: Pearson Cosine
          - type: spearman_cosine
            value: 0.8795799743051839
            name: Spearman Cosine

SentenceTransformer based on intfloat/multilingual-e5-small

This is a sentence-transformers model finetuned from intfloat/multilingual-e5-small. 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 Type: Sentence Transformer
• Base model: intfloat/multilingual-e5-small
• Maximum Sequence Length: 30 tokens
• Output Dimensionality: 384 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': 30, '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})
  (2): Normalize()
)

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("sentence_transformers_model_id")
# Run inference
sentences = [
    'child maltreatment counselling',
    'cmc',
    'book covers, flyers, posters, banners',
]
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

Semantic Similarity

• Datasets: sts-dev and sts-test
• Evaluated with EmbeddingSimilarityEvaluator

Metric	sts-dev	sts-test
pearson_cosine	0.958	0.9579
spearman_cosine	0.8789	0.8796

Training Details

Training Dataset

Unnamed Dataset

• Size: 910,013 training samples
• Columns: sentence1, sentence2, and score
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2	score
  type	string	string	float
  details	min: 3 tokens mean: 8.91 tokens max: 30 tokens	min: 3 tokens mean: 8.83 tokens max: 30 tokens	min: 0.0 mean: 0.52 max: 1.0

• Samples:

  sentence1	sentence2	score
  edición de fotografias, fondos	material selection and cognition	0.0
  professional alarm installer,service tech.,customer service relations,sales,cctv	quantity surveying & reading charts	0.1
  diagnostico ecografico	waste identification system downtime	0.19

• Loss: CosineSimilarityLoss with these parameters:

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

Evaluation Dataset

Unnamed Dataset

• Size: 113,751 evaluation samples
• Columns: sentence1, sentence2, and score
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2	score
  type	string	string	float
  details	min: 4 tokens mean: 8.89 tokens max: 30 tokens	min: 3 tokens mean: 8.96 tokens max: 30 tokens	min: 0.0 mean: 0.54 max: 1.0

• Samples:

  sentence1	sentence2	score
  a2 dutch	a2 dutch	0.98
  design of mine dumps	设计矿山废料堆	1.0
  create soil and plant improvement programmes	创建土壤和植物改良计划	1.0

• Loss: CosineSimilarityLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: epoch
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 32
• learning_rate: 1e-05
• num_train_epochs: 4
• warmup_ratio: 0.1
• fp16: True

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: epoch
• 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: 1e-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: 4
• 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}
• tp_size: 0
• 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
• 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: batch_sampler
• multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss	Validation Loss	sts-dev_spearman_cosine	sts-test_spearman_cosine
0.0352	500	0.1991	-	-	-
0.0703	1000	0.0513	-	-	-
0.1055	1500	0.0362	-	-	-
0.1407	2000	0.0331	-	-	-
0.1758	2500	0.0305	-	-	-
0.2110	3000	0.029	-	-	-
0.2461	3500	0.0273	-	-	-
0.2813	4000	0.0268	-	-	-
0.3165	4500	0.0255	-	-	-
0.3516	5000	0.0245	-	-	-
0.3868	5500	0.0238	-	-	-
0.4220	6000	0.0236	-	-	-
0.4571	6500	0.0233	-	-	-
0.4923	7000	0.0222	-	-	-
0.5275	7500	0.0225	-	-	-
0.5626	8000	0.0219	-	-	-
0.5978	8500	0.0212	-	-	-
0.6330	9000	0.0215	-	-	-
0.6681	9500	0.0207	-	-	-
0.7033	10000	0.0204	-	-	-
0.7384	10500	0.0203	-	-	-
0.7736	11000	0.0203	-	-	-
0.8088	11500	0.0202	-	-	-
0.8439	12000	0.0202	-	-	-
0.8791	12500	0.0196	-	-	-
0.9143	13000	0.0193	-	-	-
0.9494	13500	0.0193	-	-	-
0.9846	14000	0.0193	-	-	-
1.0	14219	-	0.0170	0.8694	-
1.0198	14500	0.0188	-	-	-
1.0549	15000	0.0178	-	-	-
1.0901	15500	0.0179	-	-	-
1.1253	16000	0.0178	-	-	-
1.1604	16500	0.0178	-	-	-
1.1956	17000	0.0172	-	-	-
1.2307	17500	0.0172	-	-	-
1.2659	18000	0.0175	-	-	-
1.3011	18500	0.0178	-	-	-
1.3362	19000	0.0174	-	-	-
1.3714	19500	0.0175	-	-	-
1.4066	20000	0.0171	-	-	-
1.4417	20500	0.0175	-	-	-
1.4769	21000	0.0173	-	-	-
1.5121	21500	0.0171	-	-	-
1.5472	22000	0.0174	-	-	-
1.5824	22500	0.0172	-	-	-
1.6176	23000	0.0168	-	-	-
1.6527	23500	0.0165	-	-	-
1.6879	24000	0.0169	-	-	-
1.7230	24500	0.0169	-	-	-
1.7582	25000	0.0171	-	-	-
1.7934	25500	0.0165	-	-	-
1.8285	26000	0.0165	-	-	-
1.8637	26500	0.0165	-	-	-
1.8989	27000	0.0165	-	-	-
1.9340	27500	0.0164	-	-	-
1.9692	28000	0.0164	-	-	-
2.0	28438	-	0.0153	0.8751	-
2.0044	28500	0.0162	-	-	-
2.0395	29000	0.0156	-	-	-
2.0747	29500	0.0154	-	-	-
2.1099	30000	0.0157	-	-	-
2.1450	30500	0.016	-	-	-
2.1802	31000	0.015	-	-	-
2.2153	31500	0.0155	-	-	-
2.2505	32000	0.0154	-	-	-
2.2857	32500	0.0152	-	-	-
2.3208	33000	0.0152	-	-	-
2.3560	33500	0.0152	-	-	-
2.3912	34000	0.0154	-	-	-
2.4263	34500	0.0153	-	-	-
2.4615	35000	0.0154	-	-	-
2.4967	35500	0.015	-	-	-
2.5318	36000	0.0153	-	-	-
2.5670	36500	0.0149	-	-	-
2.6022	37000	0.015	-	-	-
2.6373	37500	0.0152	-	-	-
2.6725	38000	0.0152	-	-	-
2.7076	38500	0.015	-	-	-
2.7428	39000	0.0151	-	-	-
2.7780	39500	0.0155	-	-	-
2.8131	40000	0.0148	-	-	-
2.8483	40500	0.0149	-	-	-
2.8835	41000	0.0147	-	-	-
2.9186	41500	0.015	-	-	-
2.9538	42000	0.0148	-	-	-
2.9890	42500	0.0146	-	-	-
3.0	42657	-	0.0146	0.8775	-
3.0241	43000	0.0142	-	-	-
3.0593	43500	0.0144	-	-	-
3.0945	44000	0.0146	-	-	-
3.1296	44500	0.0142	-	-	-
3.1648	45000	0.0144	-	-	-
3.1999	45500	0.0141	-	-	-
3.2351	46000	0.0142	-	-	-
3.2703	46500	0.0142	-	-	-
3.3054	47000	0.0142	-	-	-
3.3406	47500	0.0145	-	-	-
3.3758	48000	0.0142	-	-	-
3.4109	48500	0.0143	-	-	-
3.4461	49000	0.0145	-	-	-
3.4813	49500	0.0142	-	-	-
3.5164	50000	0.014	-	-	-
3.5516	50500	0.0141	-	-	-
3.5868	51000	0.0144	-	-	-
3.6219	51500	0.0143	-	-	-
3.6571	52000	0.0143	-	-	-
3.6922	52500	0.0142	-	-	-
3.7274	53000	0.014	-	-	-
3.7626	53500	0.0142	-	-	-
3.7977	54000	0.0141	-	-	-
3.8329	54500	0.0141	-	-	-
3.8681	55000	0.014	-	-	-
3.9032	55500	0.0143	-	-	-
3.9384	56000	0.0142	-	-	-
3.9736	56500	0.0141	-	-	-
4.0	56876	-	0.0146	0.8789	-
-1	-1	-	-	-	0.8796

Framework Versions

• Python: 3.11.11
• Sentence Transformers: 4.1.0
• Transformers: 4.51.3
• PyTorch: 2.6.0+cu124
• Accelerate: 1.5.2
• Datasets: 3.6.0
• Tokenizers: 0.21.1

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",
}