SentenceTransformer based on sentence-transformers/LaBSE

This is a sentence-transformers model finetuned from sentence-transformers/LaBSE. 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/LaBSE
Maximum Sequence Length: 256 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': 256, 'do_lower_case': False, 'architecture': 'BertModel'})
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Dense({'in_features': 768, 'out_features': 768, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
  (3): 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 = [
    'дуктэдеӈэ̄тын тадук вāдяӈāтын. Илӣ-ка инэӈӣдӯ Нуӈан, ариксā, илдяӈāн.',
    'и будут бить, и убьют Его: и в третий день воскреснет.',
    'Но отблеск славы будет всегда сиять на клетке, будке, конуре "Веселого": это спутник, сторож героев, свершивших во льдах удивительный подвиг.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.3893, 0.2103],
#         [0.3893, 1.0000, 0.2433],
#         [0.2103, 0.2433, 1.0000]])

Training Details

Training Dataset

Unnamed Dataset

Size: 33,487 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: 4 tokens
mean: 19.57 tokens
max: 189 tokens

min: 4 tokens
mean: 14.67 tokens
max: 149 tokens

min: 1.0
mean: 1.0
max: 1.0

	sentence_0	sentence_1	label
type	string	string	float
details	min: 4 tokens mean: 19.57 tokens max: 189 tokens	min: 4 tokens mean: 14.67 tokens max: 149 tokens	min: 1.0 mean: 1.0 max: 1.0

Samples:

sentence_0	sentence_1	label
`Да тар шаман ичичилин.`	`А этот шаман увидел.`	`1.0`
`Ляминду аманмурэн — дыгин рубле бэга.`	`У Ляминой осталась за четыре рубля в месяц.`	`1.0`
Билир-билир, элэкэс дуннэ оскедерэкин-одяракин, бэел бичэл. Егин нэкунэчэн омолгичар бидечэл. Умун тэкэчил егин куӈакар бидечэл-оскедечэл. Тыкэн биденэл, иһэвдевкил, һэгдыкэр очал. Умнэкэн Сэвэки ичэнэчэ нуӈарбатын, «Он-ке бидерэ?» — гуннэ. Эмэчэ. Тар куӈакар бэел тэкэнтын бичэл, окин-да нонон эвкил турэттэ бичэл. Эһилэ һанӈуракин, һуӈтуконди, һуӈтутоно турэһинчэл умнэт. Тыкэн эһилэ һэрэкэлтэт турэчилчэл тар куӈакар. Сэвэки тадук гунчэ:	Давным-давно, когда только что земля создавалась-появлялась, были люди. Девять братьев парнишек жили — «Единого корня» девять детей родились-проживали. Так вот живя, растут, большенькими стали. Однажды Сэвэки пошел проведать их, «Как же они живут?» — говоря. Пришел. Те дети людей корнем были, никогда до этого не разговаривали. Поэтому когда спросил (Сэвэки), по-разному, каждый по-другому (по-своему) ответили разом. Вот так теперь по отдельности (т. е. неодинаково) стали говорить те дети. Сэвэки затем сказал:	`1.0`

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim",
    "gather_across_devices": false
}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
num_train_epochs: 1
fp16: True
multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

do_predict: False
eval_strategy: steps
prediction_loss_only: True
per_device_train_batch_size: 8
per_device_eval_batch_size: 8
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: 1
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: None
warmup_ratio: None
warmup_steps: 0
log_level: passive
log_level_replica: warning
log_on_each_node: True
logging_nan_inf_filter: True
enable_jit_checkpoint: False
save_on_each_node: False
save_only_model: False
restore_callback_states_from_checkpoint: False
use_cpu: False
seed: 42
data_seed: None
bf16: False
fp16: True
bf16_full_eval: False
fp16_full_eval: False
tf32: None
local_rank: -1
ddp_backend: None
debug: []
dataloader_drop_last: False
dataloader_num_workers: 0
dataloader_prefetch_factor: None
disable_tqdm: False
remove_unused_columns: True
label_names: None
load_best_model_at_end: False
ignore_data_skip: False
fsdp: []
fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
parallelism_config: None
deepspeed: None
label_smoothing_factor: 0.0
optim: adamw_torch_fused
optim_args: None
group_by_length: False
length_column_name: length
project: huggingface
trackio_space_id: trackio
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
push_to_hub: False
resume_from_checkpoint: None
hub_model_id: None
hub_strategy: every_save
hub_private_repo: None
hub_always_push: False
hub_revision: None
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
include_for_metrics: []
eval_do_concat_batches: True
auto_find_batch_size: False
full_determinism: False
ddp_timeout: 1800
torch_compile: False
torch_compile_backend: None
torch_compile_mode: None
include_num_input_tokens_seen: no
neftune_noise_alpha: None
optim_target_modules: None
batch_eval_metrics: False
eval_on_start: False
use_liger_kernel: False
liger_kernel_config: None
eval_use_gather_object: False
average_tokens_across_devices: True
use_cache: False
prompts: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: round_robin
router_mapping: {}
learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss
0.0239	100	-
0.0478	200	-
0.0717	300	-
0.0956	400	-
0.1194	500	0.6231
0.1433	600	-
0.1672	700	-
0.1911	800	-
0.2150	900	-
0.2389	1000	0.3510
0.2628	1100	-
0.2867	1200	-
0.3106	1300	-
0.3344	1400	-
0.3583	1500	0.2851
0.3822	1600	-
0.4061	1700	-
0.4300	1800	-
0.4539	1900	-
0.4778	2000	0.1966
0.5017	2100	-
0.5256	2200	-
0.5495	2300	-
0.5733	2400	-
0.5972	2500	0.1494
0.6211	2600	-
0.6450	2700	-
0.6689	2800	-
0.6928	2900	-
0.7167	3000	0.1505
0.7406	3100	-
0.7645	3200	-

Framework Versions

Python: 3.12.12
Sentence Transformers: 5.2.2
Transformers: 5.0.0
PyTorch: 2.9.0+cu128
Accelerate: 1.12.0
Datasets: 4.0.0
Tokenizers: 0.22.2

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",
}

MultipleNegativesRankingLoss

@misc{henderson2017efficient,
    title={Efficient Natural Language Response Suggestion for Smart Reply},
    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
    year={2017},
    eprint={1705.00652},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}