metadata
language: []
library_name: sentence-transformers
tags:
- sentence-transformers
- sentence-similarity
- feature-extraction
- generated_from_trainer
- dataset_size:665
- loss:CoSENTLoss
base_model: sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2
datasets: []
metrics:
- pearson_cosine
- spearman_cosine
- pearson_manhattan
- spearman_manhattan
- pearson_euclidean
- spearman_euclidean
- pearson_dot
- spearman_dot
- pearson_max
- spearman_max
widget:
- source_sentence: Is there a free return policy?
sentences:
- general query
- faq query
- product query
- source_sentence: Quiero reservar un vuelo a Madrid
sentences:
- faq query
- general query
- product query
- source_sentence: Bestell mir einen Bluetooth-Lautsprecher
sentences:
- faq query
- general query
- general query
- source_sentence: Kann ich meinen Account auf Premium upgraden?
sentences:
- faq query
- product query
- faq query
- source_sentence: Was kostet der Versand nach Kanada?
sentences:
- product query
- faq query
- faq query
pipeline_tag: sentence-similarity
model-index:
- name: >-
SentenceTransformer based on
sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2
results:
- task:
type: semantic-similarity
name: Semantic Similarity
dataset:
name: MiniLM dev
type: MiniLM-dev
metrics:
- type: pearson_cosine
value: 0.7060858093148971
name: Pearson Cosine
- type: spearman_cosine
value: 0.7122657953703283
name: Spearman Cosine
- type: pearson_manhattan
value: 0.5850353380261794
name: Pearson Manhattan
- type: spearman_manhattan
value: 0.6010204119883696
name: Spearman Manhattan
- type: pearson_euclidean
value: 0.5997691394008732
name: Pearson Euclidean
- type: spearman_euclidean
value: 0.6079117189235353
name: Spearman Euclidean
- type: pearson_dot
value: 0.7251159526734934
name: Pearson Dot
- type: spearman_dot
value: 0.732939716175825
name: Spearman Dot
- type: pearson_max
value: 0.7251159526734934
name: Pearson Max
- type: spearman_max
value: 0.732939716175825
name: Spearman Max
- task:
type: semantic-similarity
name: Semantic Similarity
dataset:
name: MiniLM test
type: MiniLM-test
metrics:
- type: pearson_cosine
value: 0.8232712880664017
name: Pearson Cosine
- type: spearman_cosine
value: 0.822196399839697
name: Spearman Cosine
- type: pearson_manhattan
value: 0.7831863345453927
name: Pearson Manhattan
- type: spearman_manhattan
value: 0.8000293400400974
name: Spearman Manhattan
- type: pearson_euclidean
value: 0.792921493930252
name: Pearson Euclidean
- type: spearman_euclidean
value: 0.80506730817637
name: Spearman Euclidean
- type: pearson_dot
value: 0.8011854727667188
name: Pearson Dot
- type: spearman_dot
value: 0.8151432444489153
name: Spearman Dot
- type: pearson_max
value: 0.8232712880664017
name: Pearson Max
- type: spearman_max
value: 0.822196399839697
name: Spearman Max
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. 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: sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2
- Maximum Sequence Length: 128 tokens
- Output Dimensionality: 384 tokens
- 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: 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("philipp-zettl/MiniLM-similarity-small")
# Run inference
sentences = [
'Was kostet der Versand nach Kanada?',
'faq query',
'product query',
]
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
- Dataset:
MiniLM-dev
- Evaluated with
EmbeddingSimilarityEvaluator
Metric | Value |
---|---|
pearson_cosine | 0.7061 |
spearman_cosine | 0.7123 |
pearson_manhattan | 0.585 |
spearman_manhattan | 0.601 |
pearson_euclidean | 0.5998 |
spearman_euclidean | 0.6079 |
pearson_dot | 0.7251 |
spearman_dot | 0.7329 |
pearson_max | 0.7251 |
spearman_max | 0.7329 |
Semantic Similarity
- Dataset:
MiniLM-test
- Evaluated with
EmbeddingSimilarityEvaluator
Metric | Value |
---|---|
pearson_cosine | 0.8233 |
spearman_cosine | 0.8222 |
pearson_manhattan | 0.7832 |
spearman_manhattan | 0.8 |
pearson_euclidean | 0.7929 |
spearman_euclidean | 0.8051 |
pearson_dot | 0.8012 |
spearman_dot | 0.8151 |
pearson_max | 0.8233 |
spearman_max | 0.8222 |
Training Details
Training Dataset
Unnamed Dataset
- Size: 665 training samples
- Columns:
sentence1
,sentence2
, andscore
- Approximate statistics based on the first 1000 samples:
sentence1 sentence2 score type string string float details - min: 7 tokens
- mean: 11.29 tokens
- max: 19 tokens
- min: 5 tokens
- mean: 5.31 tokens
- max: 6 tokens
- min: 0.0
- mean: 0.5
- max: 1.0
- Samples:
sentence1 sentence2 score Send me deals on gaming accessories
product query
1.0
Aidez-moi à synchroniser mes contacts sur mon téléphone
faq query
0.0
Какие у вас есть предложения по ноутбукам?
faq query
0.0
- Loss:
CoSENTLoss
with these parameters:{ "scale": 20.0, "similarity_fct": "pairwise_cos_sim" }
Evaluation Dataset
Unnamed Dataset
- Size: 84 evaluation samples
- Columns:
sentence1
,sentence2
, andscore
- Approximate statistics based on the first 1000 samples:
sentence1 sentence2 score type string string float details - min: 7 tokens
- mean: 11.32 tokens
- max: 17 tokens
- min: 5 tokens
- mean: 5.42 tokens
- max: 6 tokens
- min: 0.0
- mean: 0.46
- max: 1.0
- Samples:
sentence1 sentence2 score كيف يمكنني تتبع شحنتي؟
support query
0.0
Aidez-moi à configurer une nouvelle adresse e-mail
support query
1.0
Envoyez-moi les dernières promotions sur les montres connectées
product query
1.0
- Loss:
CoSENTLoss
with these parameters:{ "scale": 20.0, "similarity_fct": "pairwise_cos_sim" }
Training Hyperparameters
Non-Default Hyperparameters
eval_strategy
: stepsper_device_train_batch_size
: 32per_device_eval_batch_size
: 32learning_rate
: 2e-05num_train_epochs
: 8warmup_ratio
: 0.1fp16
: Truebatch_sampler
: no_duplicates
All Hyperparameters
Click to expand
overwrite_output_dir
: Falsedo_predict
: Falseeval_strategy
: stepsprediction_loss_only
: Trueper_device_train_batch_size
: 32per_device_eval_batch_size
: 32per_gpu_train_batch_size
: Noneper_gpu_eval_batch_size
: Nonegradient_accumulation_steps
: 1eval_accumulation_steps
: Nonelearning_rate
: 2e-05weight_decay
: 0.0adam_beta1
: 0.9adam_beta2
: 0.999adam_epsilon
: 1e-08max_grad_norm
: 1.0num_train_epochs
: 8max_steps
: -1lr_scheduler_type
: linearlr_scheduler_kwargs
: {}warmup_ratio
: 0.1warmup_steps
: 0log_level
: passivelog_level_replica
: warninglog_on_each_node
: Truelogging_nan_inf_filter
: Truesave_safetensors
: Truesave_on_each_node
: Falsesave_only_model
: Falserestore_callback_states_from_checkpoint
: Falseno_cuda
: Falseuse_cpu
: Falseuse_mps_device
: Falseseed
: 42data_seed
: Nonejit_mode_eval
: Falseuse_ipex
: Falsebf16
: Falsefp16
: Truefp16_opt_level
: O1half_precision_backend
: autobf16_full_eval
: Falsefp16_full_eval
: Falsetf32
: Nonelocal_rank
: 0ddp_backend
: Nonetpu_num_cores
: Nonetpu_metrics_debug
: Falsedebug
: []dataloader_drop_last
: Falsedataloader_num_workers
: 0dataloader_prefetch_factor
: Nonepast_index
: -1disable_tqdm
: Falseremove_unused_columns
: Truelabel_names
: Noneload_best_model_at_end
: Falseignore_data_skip
: Falsefsdp
: []fsdp_min_num_params
: 0fsdp_config
: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}fsdp_transformer_layer_cls_to_wrap
: Noneaccelerator_config
: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}deepspeed
: Nonelabel_smoothing_factor
: 0.0optim
: adamw_torchoptim_args
: Noneadafactor
: Falsegroup_by_length
: Falselength_column_name
: lengthddp_find_unused_parameters
: Noneddp_bucket_cap_mb
: Noneddp_broadcast_buffers
: Falsedataloader_pin_memory
: Truedataloader_persistent_workers
: Falseskip_memory_metrics
: Trueuse_legacy_prediction_loop
: Falsepush_to_hub
: Falseresume_from_checkpoint
: Nonehub_model_id
: Nonehub_strategy
: every_savehub_private_repo
: Falsehub_always_push
: Falsegradient_checkpointing
: Falsegradient_checkpointing_kwargs
: Noneinclude_inputs_for_metrics
: Falseeval_do_concat_batches
: Truefp16_backend
: autopush_to_hub_model_id
: Nonepush_to_hub_organization
: Nonemp_parameters
:auto_find_batch_size
: Falsefull_determinism
: Falsetorchdynamo
: Noneray_scope
: lastddp_timeout
: 1800torch_compile
: Falsetorch_compile_backend
: Nonetorch_compile_mode
: Nonedispatch_batches
: Nonesplit_batches
: Noneinclude_tokens_per_second
: Falseinclude_num_input_tokens_seen
: Falseneftune_noise_alpha
: Noneoptim_target_modules
: Nonebatch_eval_metrics
: Falsebatch_sampler
: no_duplicatesmulti_dataset_batch_sampler
: proportional
Training Logs
Epoch | Step | Training Loss | loss | MiniLM-dev_spearman_cosine | MiniLM-test_spearman_cosine |
---|---|---|---|---|---|
0.4762 | 10 | 1.3639 | 0.8946 | 0.0665 | - |
0.9524 | 20 | 0.8488 | 0.7608 | 0.2318 | - |
1.4286 | 30 | 0.6629 | 1.0463 | 0.3736 | - |
1.9048 | 40 | 1.1413 | 1.1547 | 0.4159 | - |
2.3810 | 50 | 1.8156 | 1.2059 | 0.4760 | - |
2.8571 | 60 | 2.0179 | 0.8129 | 0.5794 | - |
3.3333 | 70 | 0.3202 | 0.6236 | 0.6217 | - |
3.8095 | 80 | 0.1437 | 0.6061 | 0.6404 | - |
4.2857 | 90 | 1.1623 | 0.7312 | 0.6424 | - |
4.7619 | 100 | 0.4376 | 0.5987 | 0.6621 | - |
5.2381 | 110 | 0.5832 | 0.4848 | 0.6837 | - |
5.7143 | 120 | 0.1749 | 0.3367 | 0.6896 | - |
6.1905 | 130 | 0.0192 | 0.2607 | 0.6936 | - |
6.6667 | 140 | 0.2047 | 0.2564 | 0.6995 | - |
7.1429 | 150 | 0.404 | 0.2747 | 0.7103 | - |
7.6190 | 160 | 0.0008 | 0.2764 | 0.7123 | - |
8.0 | 168 | - | - | - | 0.8222 |
Framework Versions
- Python: 3.10.14
- Sentence Transformers: 3.0.1
- Transformers: 4.41.2
- PyTorch: 2.3.1+cu121
- Accelerate: 0.33.0
- 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",
}
CoSENTLoss
@online{kexuefm-8847,
title={CoSENT: A more efficient sentence vector scheme than Sentence-BERT},
author={Su Jianlin},
year={2022},
month={Jan},
url={https://kexue.fm/archives/8847},
}