This is a sentence-transformers model finetuned from Alibaba-NLP/gte-multilingual-base. 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: Alibaba-NLP/gte-multilingual-base
Maximum Sequence Length: 8192 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': 8192, 'do_lower_case': False}) with Transformer model: NewModel 
  (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): 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.

python

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("sentence_transformers_model_id")
# Run inference
sentences = [
    "ليس داعشياً من بيده المسدس ..انه جندي فرنسي ينفذ اعدامات بحق مواطنين عزل في الجزائر !!! لم يكن حينها لا تنظيم قاعدة ولا دولة اسلامية ولا نصرة ليلصقوا بهم منفردين تهمة الارهاب !! انتم ام واب واخ وابن وجد الارهاب ..  Not Daashaa of the pistol in his hand .. he's a French soldier executions carried out against unarmed civilians in Algeria !!! If not then it does not regulate not base an Islamic state nor a victory for Alsqoa their individual terrorism charge !! You are a mother and father and brother and the son of terror found ..  Non Daashaa du pistolet dans sa main .. Il est un soldat français exécutions menées contre des civils non armés en Algérie !!! Si non, alors il ne réglemente pas pas fonder un Etat islamique, ni une victoire pour Alsqoa leur charge individuelle du terrorisme !! Vous êtes une mère et père et le frère et le fils de la terreur trouvé .. # occupant",
    'Massacre perpétré par des soldats français en Algérie',
    'Video Of Attack On UP Minister Shrikant Sharma',
]
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]

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: 140.38 tokens
max: 2514 tokens
min: 5 tokens
mean: 20.49 tokens
max: 141 tokens
min: 1.0
mean: 1.0
max: 1.0

Samples:

sentence_0	sentence_1	label
`Olhem aí a mineradora da Noruega destruindo o meio ambiente na Amazônia. Lula vendeu o solo para a Noruega em documento secreto. Ela arrecada 2 bilhoes ao ano e devolve 180 milhoes para consertar o estrago que ela mesmo faz na Amazônia.`	`O ex-presidente Lula vendeu o solo da Amazônia para uma empresa norueguesa`	`1.0`
EL CONGRESO DANIE Cometió una burrada Al aprobar en primera votación con 113 votos a favor, 5 en contra y una abstención, que la vacuna contra el coronavirus sea de manera OBLIGATORIA para todos Que les pasa a estos genios de la política, acaso no saben que están violando leyes universales de Derechos Humanos¿Qué les pasa a estos congresistas?. . ¿ Acaso desconocen y pisotean las leyes internacionales que respaldan los Derechos Humanos Universales ???. . Absolutamente nadie puede ser obligado a vacunarse. . Igualmente, ningún procedimiento médico puede hacerse sin el consentimiento del paciente. . No lo digo yo, lo dice la UNESCO,la Organización de las Naciones Unidas para la Educación, la Ciencia y la Cultura.... Que en sus normativas explican lo siguiente : . SOLO UNO MISMO TIENE EL CONTROL DE SU PROPIO CUERPO, nadie tiene el control de nuestro cuerpo más que uno mismo, nadie puede intervenir en nuestro cuerpo bajo ninguna circunstancia sin nuestro consentimiento. . Legalmente bajo t...	`En Perú el Congreso aprobó que la vacuna contra el covid-19 sea obligatoria`	`1.0`
`Why changes to Legislation is so difficult. Debating PTSD in Emergency Services Debating Mental Health Stigma Debating Workers Compensation Debating Cancer Legislation for Firefighters Debating MP's Pay Debating PFAS Contamination Debating Suicide Figures in Australia Debating MP's AllowancesThis tells us everything we need to know about this Government’s priorities.`	`Accurate description of photos showing the difference in attendance in various parliamentary sessions in Australia`	`1.0`

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

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 1
per_device_eval_batch_size: 1
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: 1
per_device_eval_batch_size: 1
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
0.0194	500	0.0
0.0388	1000	0.0
0.0583	1500	0.0
0.0777	2000	0.0
0.0971	2500	0.0
0.1165	3000	0.0
0.1360	3500	0.0
0.1554	4000	0.0
0.1748	4500	0.0
0.1942	5000	0.0
0.2137	5500	0.0
0.2331	6000	0.0
0.2525	6500	0.0
0.2719	7000	0.0
0.2913	7500	0.0
0.3108	8000	0.0
0.3302	8500	0.0
0.3496	9000	0.0
0.3690	9500	0.0
0.3885	10000	0.0
0.4079	10500	0.0
0.4273	11000	0.0
0.4467	11500	0.0
0.4661	12000	0.0
0.4856	12500	0.0
0.5050	13000	0.0
0.5244	13500	0.0
0.5438	14000	0.0
0.5633	14500	0.0
0.5827	15000	0.0
0.6021	15500	0.0
0.6215	16000	0.0
0.6410	16500	0.0
0.6604	17000	0.0
0.6798	17500	0.0
0.6992	18000	0.0
0.7186	18500	0.0
0.7381	19000	0.0
0.7575	19500	0.0
0.7769	20000	0.0
0.7963	20500	0.0
0.8158	21000	0.0
0.8352	21500	0.0
0.8546	22000	0.0
0.8740	22500	0.0
0.8934	23000	0.0
0.9129	23500	0.0
0.9323	24000	0.0
0.9517	24500	0.0
0.9711	25000	0.0
0.9906	25500	0.0

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