LucaZilli/all_mpnet_base_v2_190225

SentenceTransformer based on sentence-transformers/all-mpnet-base-v2

This is a sentence-transformers model finetuned from sentence-transformers/all-mpnet-base-v2. 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/all-mpnet-base-v2
• Maximum Sequence Length: 384 tokens
• Output Dimensionality: 768 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': 384, 'do_lower_case': False}) with Transformer model: MPNetModel 
  (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})
  (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("LucaZilli/all_mpnet_base_v2_190225")
# Run inference
sentences = [
    'materiali isolanti per sistemi radianti a soffitto',
    'materiali isolanti per edifici',
    'privacy and data protection training',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

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

Evaluation

Metrics

Semantic Similarity

• Datasets: custom_dataset and stsbenchmark
• Evaluated with EmbeddingSimilarityEvaluator

Metric	custom_dataset	stsbenchmark
pearson_cosine	0.7376	0.8404
spearman_cosine	0.7392	0.8342

Triplet

• Dataset: all_nli_dataset
• Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.9319

Training Details

Training Dataset

Unnamed Dataset

• Size: 25,310 training 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: 13.32 tokens max: 31 tokens	min: 4 tokens mean: 11.06 tokens max: 31 tokens	min: 0.0 mean: 0.49 max: 1.0

• Samples:

  sentence1	sentence2	score
  ottimizzazione dei tempi di produzione per capi sartoriali di lusso	strumenti per l'ottimizzazione dei tempi di produzione	0.6
  software di programmazione robotica per lucidatura	software gestionale generico	0.4
  rete di sensori per l'analisi del suolo in tempo reale	software per gestione aziendale	0.0

• Loss: CosineSimilarityLoss with these parameters:

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

Evaluation Dataset

Unnamed Dataset

• Size: 3,164 evaluation samples
• Columns: sentence1, sentence2, and score
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2	score
  type	string	string	float
  details	min: 5 tokens mean: 13.61 tokens max: 31 tokens	min: 4 tokens mean: 11.39 tokens max: 27 tokens	min: 0.0 mean: 0.49 max: 1.0

• Samples:

  sentence1	sentence2	score
  ispezioni regolari per camion aziendali	ispezioni regolari per camion di consegna	1.0
  blister packaging machines GMP compliant	food packaging machines	0.4
  EMI shielding paints for electronics	Vernici per schermatura elettromagnetica dispositivi elettronici	0.8

• Loss: CosineSimilarityLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• num_train_epochs: 5
• 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: 16
• per_device_eval_batch_size: 16
• 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: 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: 5
• 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: 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
• 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
• use_liger_kernel: False
• eval_use_gather_object: False
• average_tokens_across_devices: False
• prompts: None
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	Validation Loss	custom_dataset_spearman_cosine	all_nli_dataset_cosine_accuracy	stsbenchmark_spearman_cosine
-1	-1	-	-	0.7392	0.9319	0.8342
0.0632	100	0.0484	-	-	-	-
0.1264	200	0.0389	0.0363	-	-	-
0.1896	300	0.0377	-	-	-	-
0.2528	400	0.033	0.0307	-	-	-
0.3161	500	0.0302	-	-	-	-
0.3793	600	0.0329	0.0298	-	-	-
0.4425	700	0.0315	-	-	-	-
0.5057	800	0.0308	0.0275	-	-	-
0.5689	900	0.0309	-	-	-	-
0.6321	1000	0.0279	0.0282	-	-	-
0.6953	1100	0.0299	-	-	-	-
0.7585	1200	0.0286	0.0271	-	-	-
0.8217	1300	0.0292	-	-	-	-
0.8850	1400	0.027	0.0261	-	-	-
0.9482	1500	0.0254	-	-	-	-
1.0114	1600	0.0237	0.0242	-	-	-
1.0746	1700	0.0196	-	-	-	-
1.1378	1800	0.0196	0.0242	-	-	-
1.2010	1900	0.021	-	-	-	-
1.2642	2000	0.0225	0.0267	-	-	-
1.3274	2100	0.022	-	-	-	-
1.3906	2200	0.0203	0.0246	-	-	-
1.4539	2300	0.0178	-	-	-	-
1.5171	2400	0.019	0.0240	-	-	-
1.5803	2500	0.0209	-	-	-	-
1.6435	2600	0.0186	0.0240	-	-	-
1.7067	2700	0.0261	-	-	-	-
1.7699	2800	0.0193	0.0246	-	-	-
1.8331	2900	0.02	-	-	-	-
1.8963	3000	0.0196	0.0240	-	-	-
1.9595	3100	0.0186	-	-	-	-
2.0228	3200	0.0164	0.0226	-	-	-
2.0860	3300	0.0122	-	-	-	-
2.1492	3400	0.0123	0.0221	-	-	-
2.2124	3500	0.0135	-	-	-	-
2.2756	3600	0.0134	0.0226	-	-	-
2.3388	3700	0.0128	-	-	-	-
2.4020	3800	0.0126	0.0231	-	-	-
2.4652	3900	0.0134	-	-	-	-
2.5284	4000	0.0142	0.0231	-	-	-
2.5917	4100	0.0124	-	-	-	-
2.6549	4200	0.0132	0.0215	-	-	-
2.7181	4300	0.0136	-	-	-	-
2.7813	4400	0.013	0.0218	-	-	-
2.8445	4500	0.0127	-	-	-	-
2.9077	4600	0.0126	0.0213	-	-	-
2.9709	4700	0.0133	-	-	-	-
3.0341	4800	0.0103	0.0209	-	-	-
3.0973	4900	0.0086	-	-	-	-
3.1606	5000	0.0088	0.0211	-	-	-
3.2238	5100	0.0081	-	-	-	-
3.2870	5200	0.0079	0.0212	-	-	-
3.3502	5300	0.0094	-	-	-	-
3.4134	5400	0.0086	0.0210	-	-	-
3.4766	5500	0.0089	-	-	-	-
3.5398	5600	0.009	0.0209	-	-	-
3.6030	5700	0.0086	-	-	-	-
3.6662	5800	0.0087	0.0213	-	-	-
3.7295	5900	0.0085	-	-	-	-
3.7927	6000	0.0094	0.0213	-	-	-
3.8559	6100	0.0095	-	-	-	-
3.9191	6200	0.01	0.0212	-	-	-
3.9823	6300	0.0097	-	-	-	-
4.0455	6400	0.0081	0.0214	-	-	-
4.1087	6500	0.0095	-	-	-	-
4.1719	6600	0.0083	0.0208	-	-	-
4.2351	6700	0.0082	-	-	-	-
4.2984	6800	0.0074	0.0208	-	-	-
4.3616	6900	0.0076	-	-	-	-
4.4248	7000	0.0072	0.0205	-	-	-
4.4880	7100	0.0072	-	-	-	-
4.5512	7200	0.007	0.0205	-	-	-
4.6144	7300	0.0069	-	-	-	-
4.6776	7400	0.0069	0.0203	-	-	-
4.7408	7500	0.0068	-	-	-	-
4.8040	7600	0.0073	0.0204	-	-	-
4.8673	7700	0.0063	-	-	-	-
4.9305	7800	0.0065	0.0203	-	-	-
4.9937	7900	0.0069	-	-	-	-

Framework Versions

• Python: 3.11.11
• Sentence Transformers: 3.4.1
• Transformers: 4.48.3
• PyTorch: 2.5.1+cu124
• Accelerate: 1.3.0
• Datasets: 3.3.1
• Tokenizers: 0.21.0

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",
}