AraBERT v2 base trained on Arabic triplets
This is a sentence-transformers model finetuned from aubmindlab/bert-base-arabertv02. 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: aubmindlab/bert-base-arabertv02
- Maximum Sequence Length: 128 tokens
- Output Dimensionality: 768 dimensions
- Similarity Function: Cosine Similarity
- Language: ar
- License: apache-2.0
Model Sources
Full Model Architecture
SentenceTransformer(
(0): Transformer({'max_seq_length': 128, 'do_lower_case': False, 'architecture': 'BertModel'})
(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})
)
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
model = SentenceTransformer("hindalmayyali/Sentence_arabertV1")
sentences = [
'ما الفرق بين الحذف والتقطيع؟',
'ما هي الاختلافات بين الحذف والتقطيع؟',
'أي كتاب أفضل لـ (نيت) ؟',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
similarities = model.similarity(embeddings, embeddings)
print(similarities)
Evaluation
Metrics
Triplet
| Metric |
arabic-valid |
arabic-test |
| cosine_accuracy |
0.921 |
0.9328 |
Training Details
Training Dataset
Unnamed Dataset
Evaluation Dataset
Unnamed Dataset
Training Hyperparameters
Non-Default Hyperparameters
eval_strategy: epoch
per_device_train_batch_size: 32
per_device_eval_batch_size: 48
gradient_accumulation_steps: 2
learning_rate: 2e-05
num_train_epochs: 10
warmup_ratio: 0.1
load_best_model_at_end: True
optim: adamw_torch
dataloader_pin_memory: False
gradient_checkpointing: True
batch_sampler: no_duplicates
All Hyperparameters
Click to expand
overwrite_output_dir: False
do_predict: False
eval_strategy: epoch
prediction_loss_only: True
per_device_train_batch_size: 32
per_device_eval_batch_size: 48
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 2
eval_accumulation_steps: None
torch_empty_cache_steps: None
learning_rate: 2e-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: 10
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: 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: True
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}
parallelism_config: 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: False
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
hub_revision: None
gradient_checkpointing: True
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
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
liger_kernel_config: None
eval_use_gather_object: False
average_tokens_across_devices: False
prompts: None
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional
router_mapping: {}
learning_rate_mapping: {}
Training Logs
Click to expand
| Epoch |
Step |
Training Loss |
Validation Loss |
arabic-valid_cosine_accuracy |
arabic-test_cosine_accuracy |
| 0.0013 |
1 |
2.3706 |
- |
- |
- |
| 0.0640 |
50 |
2.0507 |
- |
- |
- |
| 0.1280 |
100 |
1.2702 |
- |
- |
- |
| 0.1919 |
150 |
0.8272 |
- |
- |
- |
| 0.2559 |
200 |
0.665 |
- |
- |
- |
| 0.3199 |
250 |
0.612 |
- |
- |
- |
| 0.3839 |
300 |
0.5628 |
- |
- |
- |
| 0.4479 |
350 |
0.5455 |
- |
- |
- |
| 0.5118 |
400 |
0.484 |
- |
- |
- |
| 0.5758 |
450 |
0.4632 |
- |
- |
- |
| 0.6398 |
500 |
0.4618 |
- |
- |
- |
| 0.7038 |
550 |
0.4384 |
- |
- |
- |
| 0.7678 |
600 |
0.4087 |
- |
- |
- |
| 0.8317 |
650 |
0.3721 |
- |
- |
- |
| 0.8957 |
700 |
0.4261 |
- |
- |
- |
| 0.9597 |
750 |
0.4114 |
- |
- |
- |
| 1.0 |
782 |
- |
0.4069 |
0.9200 |
- |
| 1.0230 |
800 |
0.3637 |
- |
- |
- |
| 1.0870 |
850 |
0.3329 |
- |
- |
- |
| 1.1510 |
900 |
0.3133 |
- |
- |
- |
| 1.2150 |
950 |
0.2974 |
- |
- |
- |
| 1.2790 |
1000 |
0.2944 |
- |
- |
- |
| 1.3429 |
1050 |
0.2627 |
- |
- |
- |
| 1.4069 |
1100 |
0.2994 |
- |
- |
- |
| 1.4709 |
1150 |
0.3068 |
- |
- |
- |
| 1.5349 |
1200 |
0.3016 |
- |
- |
- |
| 1.5988 |
1250 |
0.3068 |
- |
- |
- |
| 1.6628 |
1300 |
0.304 |
- |
- |
- |
| 1.7268 |
1350 |
0.3012 |
- |
- |
- |
| 1.7908 |
1400 |
0.2998 |
- |
- |
- |
| 1.8548 |
1450 |
0.3181 |
- |
- |
- |
| 1.9187 |
1500 |
0.2858 |
- |
- |
- |
| 1.9827 |
1550 |
0.2843 |
- |
- |
- |
| 2.0 |
1564 |
- |
0.3489 |
0.9340 |
- |
| 2.0461 |
1600 |
0.223 |
- |
- |
- |
| 2.1100 |
1650 |
0.191 |
- |
- |
- |
| 2.1740 |
1700 |
0.1828 |
- |
- |
- |
| 2.2380 |
1750 |
0.1762 |
- |
- |
- |
| 2.3020 |
1800 |
0.2021 |
- |
- |
- |
| 2.3660 |
1850 |
0.1824 |
- |
- |
- |
| 2.4299 |
1900 |
0.1873 |
- |
- |
- |
| 2.4939 |
1950 |
0.188 |
- |
- |
- |
| 2.5579 |
2000 |
0.188 |
- |
- |
- |
| 2.6219 |
2050 |
0.1909 |
- |
- |
- |
| 2.6859 |
2100 |
0.1888 |
- |
- |
- |
| 2.7498 |
2150 |
0.1839 |
- |
- |
- |
| 2.8138 |
2200 |
0.1965 |
- |
- |
- |
| 2.8778 |
2250 |
0.1881 |
- |
- |
- |
| 2.9418 |
2300 |
0.1753 |
- |
- |
- |
| 3.0 |
2346 |
- |
0.3480 |
0.9290 |
- |
| 3.0051 |
2350 |
0.1847 |
- |
- |
- |
| 3.0691 |
2400 |
0.1415 |
- |
- |
- |
| 3.1331 |
2450 |
0.1258 |
- |
- |
- |
| 3.1971 |
2500 |
0.1125 |
- |
- |
- |
| 3.2610 |
2550 |
0.1186 |
- |
- |
- |
| 3.3250 |
2600 |
0.1235 |
- |
- |
- |
| 3.3890 |
2650 |
0.1328 |
- |
- |
- |
| 3.4530 |
2700 |
0.1294 |
- |
- |
- |
| 3.5170 |
2750 |
0.138 |
- |
- |
- |
| 3.5809 |
2800 |
0.1282 |
- |
- |
- |
| 3.6449 |
2850 |
0.1391 |
- |
- |
- |
| 3.7089 |
2900 |
0.1321 |
- |
- |
- |
| 3.7729 |
2950 |
0.1396 |
- |
- |
- |
| 3.8369 |
3000 |
0.1344 |
- |
- |
- |
| 3.9008 |
3050 |
0.1257 |
- |
- |
- |
| 3.9648 |
3100 |
0.1441 |
- |
- |
- |
| 4.0 |
3128 |
- |
0.3466 |
0.924 |
- |
| 4.0282 |
3150 |
0.1105 |
- |
- |
- |
| 4.0921 |
3200 |
0.0954 |
- |
- |
- |
| 4.1561 |
3250 |
0.0894 |
- |
- |
- |
| 4.2201 |
3300 |
0.0945 |
- |
- |
- |
| 4.2841 |
3350 |
0.0958 |
- |
- |
- |
| 4.3480 |
3400 |
0.0957 |
- |
- |
- |
| 4.4120 |
3450 |
0.0935 |
- |
- |
- |
| 4.4760 |
3500 |
0.1093 |
- |
- |
- |
| 4.5400 |
3550 |
0.1107 |
- |
- |
- |
| 4.6040 |
3600 |
0.0995 |
- |
- |
- |
| 4.6679 |
3650 |
0.1081 |
- |
- |
- |
| 4.7319 |
3700 |
0.0887 |
- |
- |
- |
| 4.7959 |
3750 |
0.0952 |
- |
- |
- |
| 4.8599 |
3800 |
0.0976 |
- |
- |
- |
| 4.9239 |
3850 |
0.1034 |
- |
- |
- |
| 4.9878 |
3900 |
0.0903 |
- |
- |
- |
| 5.0 |
3910 |
- |
0.3495 |
0.9240 |
- |
| 5.0512 |
3950 |
0.0748 |
- |
- |
- |
| 5.1152 |
4000 |
0.0881 |
- |
- |
- |
| 5.1791 |
4050 |
0.0721 |
- |
- |
- |
| 5.2431 |
4100 |
0.0811 |
- |
- |
- |
| 5.3071 |
4150 |
0.0834 |
- |
- |
- |
| 5.3711 |
4200 |
0.0936 |
- |
- |
- |
| 5.4351 |
4250 |
0.0769 |
- |
- |
- |
| 5.4990 |
4300 |
0.0817 |
- |
- |
- |
| 5.5630 |
4350 |
0.078 |
- |
- |
- |
| 5.6270 |
4400 |
0.0925 |
- |
- |
- |
| 5.6910 |
4450 |
0.0773 |
- |
- |
- |
| 5.7550 |
4500 |
0.0779 |
- |
- |
- |
| 5.8189 |
4550 |
0.0726 |
- |
- |
- |
| 5.8829 |
4600 |
0.0864 |
- |
- |
- |
| 5.9469 |
4650 |
0.0779 |
- |
- |
- |
| 6.0 |
4692 |
- |
0.3469 |
0.9250 |
- |
| 6.0102 |
4700 |
0.0795 |
- |
- |
- |
| 6.0742 |
4750 |
0.0673 |
- |
- |
- |
| 6.1382 |
4800 |
0.0653 |
- |
- |
- |
| 6.2022 |
4850 |
0.0638 |
- |
- |
- |
| 6.2662 |
4900 |
0.0597 |
- |
- |
- |
| 6.3301 |
4950 |
0.0705 |
- |
- |
- |
| 6.3941 |
5000 |
0.0664 |
- |
- |
- |
| 6.4581 |
5050 |
0.0675 |
- |
- |
- |
| 6.5221 |
5100 |
0.0742 |
- |
- |
- |
| 6.5861 |
5150 |
0.0704 |
- |
- |
- |
| 6.6500 |
5200 |
0.06 |
- |
- |
- |
| 6.7140 |
5250 |
0.0586 |
- |
- |
- |
| 6.7780 |
5300 |
0.0643 |
- |
- |
- |
| 6.8420 |
5350 |
0.0699 |
- |
- |
- |
| 6.9060 |
5400 |
0.067 |
- |
- |
- |
| 6.9699 |
5450 |
0.0643 |
- |
- |
- |
| 7.0 |
5474 |
- |
0.3491 |
0.9210 |
- |
| -1 |
-1 |
- |
- |
- |
0.9328 |
- The bold row denotes the saved checkpoint.
Framework Versions
- Python: 3.13.2
- Sentence Transformers: 5.1.0
- Transformers: 4.56.1
- PyTorch: 2.8.0
- Accelerate: 1.10.1
- Datasets: 4.0.0
- Tokenizers: 0.22.0
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",
}
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}
}