SentenceTransformer based on sentence-transformers/all-MiniLM-L6-v2

This is a sentence-transformers model finetuned from sentence-transformers/all-MiniLM-L6-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/all-MiniLM-L6-v2
Maximum Sequence Length: 256 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': 256, '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 = [
    'indulge your senses with comme une evidence limited edition 2008 by yves rocher a sophisticated floral fragrance that captures the essence of tranquility and elegance this scent harmoniously blends delicate floral notes with hints of earthy moss creating a fresh and uplifting experience reminiscent of a serene garden in full bloom users describe it as both refreshing and subtle ideal for those seeking a signature scent that exudes femininity without overwhelming presence the composition is said to invoke feelings of serenity and poise making it perfect for daytime wear or special occasions when one desires a touch of grace with an overall rating of 376 the fragrance has garnered appreciation for its longevity and ability to evoke memories of blooming florals intertwined with natural sweetness it strikes a perfect balance appealing to those who cherish a scent that is both light and intricately layered whether strolling through sunlit paths or enjoying quiet moments inside comme une evidence limited edition envelops the wearer in a soothing embrace leaving a lasting impression of refined simplicity',
    'oud, ginger, sea salt, lily, resins',
    'papyrus, ginger, spices, herbal notes, lemon blossom, green tree accord, ambertonic, lemon leaf oil, cassis, pimento, acacia, citron, gardenia, elemi, black amber, clove, clary sage, ambergris, lime, darjeeling tea, cashmeran, blonde woods',
]
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

Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.934
spearman_cosine	0.7334

Training Details

Training Dataset

Unnamed Dataset

Size: 1,459 training samples
Columns: sentence_0, sentence_1, and label
Approximate statistics based on the first 1000 samples:
sentence_0 sentence_1 label
type string string float
details
min: 12 tokens
mean: 182.01 tokens
max: 256 tokens

min: 5 tokens
mean: 33.09 tokens
max: 88 tokens

min: 0.0
mean: 0.25
max: 1.0

	sentence_0	sentence_1	label
type	string	string	float
details	min: 12 tokens mean: 182.01 tokens max: 256 tokens	min: 5 tokens mean: 33.09 tokens max: 88 tokens	min: 0.0 mean: 0.25 max: 1.0

Samples:

sentence_0	sentence_1	label
today tomorrow always in love by avon embodying a sense of timeless romance today tomorrow always in love is an enchanting fragrance that strikes a perfect balance between freshness and warmth this captivating scent opens with bright effervescent notes that evoke images of blooming gardens and sunlit moments as the fragrance unfolds it reveals a charming bouquet that celebrates femininity featuring delicate floral elements that wrap around the wearer like a cherished embrace users describe this perfume as uplifting and evocative making it an ideal companion for both everyday wear and special occasions many reviewers appreciate its elegant character highlighting its multifaceted nature that seamlessly transitions from day to night while some find it subtly sweet and playful others cherish its musky undertones which lend a depth that enhances its allure overall with a moderate rating that suggests a solid appreciation among wearers today tomorrow always in love captures the essence of ro...	`lotus, neroli, carambola, pomegranate, tuberose, gardenia, tuberose, pepper, musk, woody notes, amber`	`1.0`
mankind hero by kenneth cole encapsulates a vibrant and adventurous spirit designed for the modern man who embraces both freshness and sophistication this fragrance unfolds with an invigorating burst reminiscent of a brisk mountain breeze seamlessly paired with a zesty hint of citrus the aromatic heart introduces a soothing edginess where lavender and warm vanilla intertwine creating a balanced yet captivating profile as it settles an inviting warmth emerges enriched by woody undertones that linger pleasantly on the skin users have praised mankind hero for its versatile character suitable for both casual outings and formal occasions many describe it as longlasting and unique appreciating the balanced blend that feels both refreshing and comforting the overall sentiment reflects a sense of confidence and elegance making this scent a cherished addition to a mans fragrance collection it has garnered favorable reviews boasting a solid rating that underscores its appeal embrace the essence ...	`mountain air, lemon, coriander, lavender, vanilla, clary sage, plum, musk, coumarin, amberwood, oak moss`	`1.0`
black essential dark by avon immerse yourself in the captivating allure of black essential dark a fragrance that elegantly marries the depth of aromatic woods with a touch of leathers sensuality this modern scent envelops the wearer in a rich and sophisticated aura exuding confidence and a hint of mystery users describe it as both refreshing and spicy with an invigorating blend that feels perfect for the urban man who embraces lifes more daring adventures crafted with meticulous attention by perfumer mike parrot this fragrance has garnered a solid reputation amongst enthusiasts resulting in a commendable 405 rating from its admirers many find it to be versatile enough for both day and night wear making it an essential companion for various occasions reviewers frequently highlight its longlasting presence creating an inviting and memorable impression with a delicate yet commanding presence black essential dark is ideal for those looking to leave a mark without overpowering the senses wh...	`mint, allspice, white tea, amber, herbal notes, pear blossom, armoise, gurgum wood, creme brulee`	`0.0`

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: 32
per_device_eval_batch_size: 32
num_train_epochs: 5
multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
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: 5e-05
weight_decay: 0.0
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
max_grad_norm: 1
num_train_epochs: 5
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: {}
warmup_ratio: 0.0
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: False
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: batch_sampler
multi_dataset_batch_sampler: round_robin

Training Logs

Epoch	Step	spearman_cosine
1.0	46	0.6586
1.0870	50	0.6783
2.0	92	0.7334
2.1739	100	0.7268
3.0	138	0.7400
3.2609	150	0.7400
4.0	184	0.7426
4.3478	200	0.7387
5.0	230	0.7400
1.0	46	0.7387
1.0870	50	0.7387
2.0	92	0.7295
2.1739	100	0.7255
3.0	138	0.7242
3.2609	150	0.7255
4.0	184	0.7124
4.3478	200	0.7216
5.0	230	0.7334

Framework Versions

Python: 3.11.11
Sentence Transformers: 3.3.1
Transformers: 4.47.1
PyTorch: 2.5.1+cu124
Accelerate: 1.2.1
Datasets: 3.2.0
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",
}

dawn78
/

minilm6_perfumerecommender_v3