This is a sentence-transformers model finetuned from Lajavaness/bilingual-embedding-small. 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: Lajavaness/bilingual-embedding-small
Maximum Sequence Length: 512 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': 512, 'do_lower_case': False}) with Transformer model: BilingualModel 
  (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.

python

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("sentence_transformers_model_id")
# Run inference
sentences = [
    'This is how the buildings moved this morning in Mexico City',
    'The video captures the earthquake of June 23, 2020',
    'Photo shows bombing in Gaza in January 2022',
]
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]

Training Details

Training Dataset

Unnamed Dataset

Size: 25,743 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: 2 tokens
mean: 110.17 tokens
max: 512 tokens
min: 6 tokens
mean: 19.45 tokens
max: 190 tokens
min: 1.0
mean: 1.0
max: 1.0

Samples:

sentence_0	sentence_1	label
best music k.m KOSE CELLIE HINS GUINOT SKIN CARE KWhat people fear most is not being physically disabled, but giving up on themselves. There are still many beautiful things in life to aspire to! This stunning performance, known as the American spirit, brought tears to the eyes of 10,000 spectators. Male dancer Babo has been blind since childhood due to a fire in his home. In order to protect him, his mother held him tightly in her arms and jumped from the 7th floor. The mother died as a result, and the little baby became blind due to bleeding from the fundus. His mother was an ice skater before he died, and Babo also had a soft spot for ice skating. Although he couldn't see anything, he still pursued dance enthusiastically. He danced the famous tango "La Cumparsita" with his partner at the World Figure Skating Championships in Helsinki! 1. His ears are like bats that can measure the sound and distance around him. 2. The female dancer is very amazing. She danced with him and led him for...	`Performance by a blind American ice dancer`	`1.0`
`Photo from 2016. "Good" times when health was "fine" and the press did not report anything about. Bunch of Hypocrites...Let's go fight my people... . left right not army above all`	`Photo of a hospital in 2016. Good times when health was "good" and the press didn't report anything about it`	`1.0`
`Haifa Oh Tel Aviv-Yafo Oh N WEST BANK Jerusalem is GAZA STRIPE Be'er Sheva Israel 65 65 35 35 15 M5 10 40Google and Apple maps have officially removed Palestine from the World Maps. Today Palestine was erased from the maps tomorrow Palestine will be erased from the world. PUT PALESTINE BACK ON THE MAP. Please unite now Pakistanio. Enemy is very strong if we are divided. Think just about Pakistan. Support each other, support Pakistan and support your leadership.`	`Google and Apple removed Palestine from its maps`	`1.0`

Loss: MultipleNegativesRankingLoss with these parameters:
json
1{ 2 "scale": 20.0, 3 "similarity_fct": "cos_sim" 4}

Training Hyperparameters

Non-Default Hyperparameters

num_train_epochs: 1
multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: no
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
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: {}
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	Training Loss
0.1554	500	0.1021
0.3108	1000	0.0732
0.4661	1500	0.0781
0.6215	2000	0.0762
0.7769	2500	0.0763
0.9323	3000	0.0739
0.1554	500	0.0474
0.3108	1000	0.0478
0.4661	1500	0.0558
0.6215	2000	0.0542
0.7769	2500	0.0457
0.9323	3000	0.0395
0.1554	500	0.0055
0.3108	1000	0.0046
0.4661	1500	0.0099
0.6215	2000	0.0157
0.7769	2500	0.0176
0.9323	3000	0.0257

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

bibtex

@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