philipp-zettl/MiniLM-similarity-small

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 = [
    'Envoyez-moi la politique de garantie de ce produit',
    'faq query',
    'account 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	0.6538
spearman_cosine	0.6337
pearson_manhattan	0.58
spearman_manhattan	0.5526
pearson_euclidean	0.5732
spearman_euclidean	0.5395
pearson_dot	0.636
spearman_dot	0.6238
pearson_max	0.6538
spearman_max	0.6337

Semantic Similarity

• Dataset: MiniLM-test
• Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.6682
spearman_cosine	0.6222
pearson_manhattan	0.5715
spearman_manhattan	0.5481
pearson_euclidean	0.5727
spearman_euclidean	0.5493
pearson_dot	0.6396
spearman_dot	0.6107
pearson_max	0.6682
spearman_max	0.6222

Training Details

Training Dataset

Unnamed Dataset

• Size: 1,267 training samples
• Columns: sentence1, sentence2, and score
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2	score
  type	string	string	float
  details	min: 6 tokens mean: 10.77 tokens max: 18 tokens	min: 4 tokens mean: 5.31 tokens max: 6 tokens	min: 0.0 mean: 0.67 max: 1.0

• Samples:

  sentence1	sentence2	score
  Get information on the next art exhibition	product query	0.0
  Show me how to update my profile	product query	0.0
  Покажите мне доступные варианты полетов в Турцию	faq query	0.0

• Loss: CoSENTLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "pairwise_cos_sim"
  }
  

Evaluation Dataset

Unnamed Dataset

• Size: 159 evaluation samples
• Columns: sentence1, sentence2, and score
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2	score
  type	string	string	float
  details	min: 6 tokens mean: 10.65 tokens max: 17 tokens	min: 4 tokens mean: 5.35 tokens max: 6 tokens	min: 0.0 mean: 0.67 max: 1.0

• Samples:

  sentence1	sentence2	score
  Sende mir die Bestellbestätigung per E-Mail	order query	0.0
  How do I add a new payment method?	faq query	1.0
  No puedo conectar mi impresora, ¿puedes ayudarme?	support query	1.0

• Loss: CoSENTLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "pairwise_cos_sim"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• learning_rate: 2e-05
• num_train_epochs: 2
• 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: 8
• per_device_eval_batch_size: 8
• 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: 2
• 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
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	loss	MiniLM-dev_spearman_cosine	MiniLM-test_spearman_cosine
0.0629	10	6.2479	2.5890	0.1448	-
0.1258	20	4.3549	2.2787	0.1965	-
0.1887	30	3.5969	2.0104	0.2599	-
0.2516	40	2.4979	1.7269	0.3357	-
0.3145	50	2.5551	1.5747	0.4439	-
0.3774	60	3.1446	1.4892	0.4750	-
0.4403	70	2.1353	1.5305	0.4662	-
0.5031	80	2.9341	1.3718	0.4848	-
0.5660	90	2.8709	1.2469	0.5316	-
0.6289	100	2.1367	1.2558	0.5436	-
0.6918	110	2.2735	1.2939	0.5392	-
0.7547	120	2.8646	1.1206	0.5616	-
0.8176	130	3.3204	1.0213	0.5662	-
0.8805	140	0.8989	0.9866	0.5738	-
0.9434	150	0.0057	0.9961	0.5674	-
1.0063	160	0.0019	1.0111	0.5674	-
1.0692	170	0.4617	1.0275	0.5747	-
1.1321	180	0.0083	1.0746	0.5732	-
1.1950	190	0.5048	1.0968	0.5753	-
1.2579	200	0.0002	1.0840	0.5738	-
1.3208	210	0.07	1.0364	0.5753	-
1.3836	220	0.0	0.9952	0.5750	-
1.4465	230	0.0	0.9922	0.5744	-
1.5094	240	0.0	0.9923	0.5726	-
1.0126	250	0.229	0.9930	0.5729	-
1.0755	260	2.2061	0.9435	0.5880	-
1.1384	270	2.7711	0.8892	0.6078	-
1.2013	280	0.7528	0.8886	0.6148	-
1.2642	290	0.386	0.8927	0.6162	-
1.3270	300	0.8902	0.8710	0.6267	-
1.3899	310	0.9534	0.8429	0.6337	-
1.4403	318	-	-	-	0.6222

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},
}