Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks
Paper
•
1908.10084
•
Published
•
10
This is a sentence-transformers model finetuned from dangvantuan/vietnamese-document-embedding. 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.
SentenceTransformer(
(0): Transformer({'max_seq_length': 8192, 'do_lower_case': False, 'architecture': 'VietnameseModel'})
(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()
)
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("TTHDZ/finetuned_vietnamese-document-embedding")
# Run inference
sentences = [
'Thách thức nào đặc thù khi huấn luyện AI cho an ninh mạng?',
'Public_076\nKiến trúc tổng thể của hệ thống phòng thủ AI\nTầng phân tích và học máy\n* Giám sát (Supervised Learning): sử dụng dữ liệu đã gắn nhãn (ví dụ, gói tin tấn công) để dự đoán tấn công đã biết.\n * Không giám sát (Unsupervised/Anomaly Detection): tìm kiếm mẫu hành vi bất thường, hữu ích với các cuộc tấn công 0-day.\n * Học bán giám sát và tự giám sát: giảm phụ thuộc vào dữ liệu gắn nhãn khan hiếm.\n * Học tăng cường (Reinforcement Learning): cho phép hệ thống tự điều chỉnh chính sách phản ứng dựa trên kết quả.',
'Public_076\nKiến trúc tổng thể của hệ thống phòng thủ AI\nMột giải pháp AI an ninh mạng toàn diện thường bao gồm nhiều lớp:',
]
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., 1., 1.],
# [1., 1., 1.],
# [1., 1., 1.]])
sentence_0, sentence_1, and sentence_2| sentence_0 | sentence_1 | sentence_2 | |
|---|---|---|---|
| type | string | string | string |
| details |
|
|
|
| sentence_0 | sentence_1 | sentence_2 |
|---|---|---|
Theo tài liệu Public_503, bộ kí tự (character set) trong ngôn ngữ lập trình có chức năng gì? |
Public_503 |
Public_183 |
Theo tài liệu Public_194, khái niệm Database trong MongoDB là gì? |
Public_194 |
Public_194 |
Dự án Stellarator nào được đề cập trong tài liệu? |
Public_096 |
Public_096 |
TripletLoss with these parameters:{
"distance_metric": "TripletDistanceMetric.EUCLIDEAN",
"triplet_margin": 5
}
per_device_train_batch_size: 2per_device_eval_batch_size: 2fp16: Truemulti_dataset_batch_sampler: round_robinoverwrite_output_dir: Falsedo_predict: Falseeval_strategy: noprediction_loss_only: Trueper_device_train_batch_size: 2per_device_eval_batch_size: 2per_gpu_train_batch_size: Noneper_gpu_eval_batch_size: Nonegradient_accumulation_steps: 1eval_accumulation_steps: Nonetorch_empty_cache_steps: Nonelearning_rate: 5e-05weight_decay: 0.0adam_beta1: 0.9adam_beta2: 0.999adam_epsilon: 1e-08max_grad_norm: 1num_train_epochs: 3max_steps: -1lr_scheduler_type: linearlr_scheduler_kwargs: {}warmup_ratio: 0.0warmup_steps: 0log_level: passivelog_level_replica: warninglog_on_each_node: Truelogging_nan_inf_filter: Truesave_safetensors: Truesave_on_each_node: Falsesave_only_model: Falserestore_callback_states_from_checkpoint: Falseno_cuda: Falseuse_cpu: Falseuse_mps_device: Falseseed: 42data_seed: Nonejit_mode_eval: Falsebf16: Falsefp16: Truefp16_opt_level: O1half_precision_backend: autobf16_full_eval: Falsefp16_full_eval: Falsetf32: Nonelocal_rank: 0ddp_backend: Nonetpu_num_cores: Nonetpu_metrics_debug: Falsedebug: []dataloader_drop_last: Falsedataloader_num_workers: 0dataloader_prefetch_factor: Nonepast_index: -1disable_tqdm: Falseremove_unused_columns: Truelabel_names: Noneload_best_model_at_end: Falseignore_data_skip: Falsefsdp: []fsdp_min_num_params: 0fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}fsdp_transformer_layer_cls_to_wrap: Noneaccelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}parallelism_config: Nonedeepspeed: Nonelabel_smoothing_factor: 0.0optim: adamw_torchoptim_args: Noneadafactor: Falsegroup_by_length: Falselength_column_name: lengthproject: huggingfacetrackio_space_id: trackioddp_find_unused_parameters: Noneddp_bucket_cap_mb: Noneddp_broadcast_buffers: Falsedataloader_pin_memory: Truedataloader_persistent_workers: Falseskip_memory_metrics: Trueuse_legacy_prediction_loop: Falsepush_to_hub: Falseresume_from_checkpoint: Nonehub_model_id: Nonehub_strategy: every_savehub_private_repo: Nonehub_always_push: Falsehub_revision: Nonegradient_checkpointing: Falsegradient_checkpointing_kwargs: Noneinclude_inputs_for_metrics: Falseinclude_for_metrics: []eval_do_concat_batches: Truefp16_backend: autopush_to_hub_model_id: Nonepush_to_hub_organization: Nonemp_parameters: auto_find_batch_size: Falsefull_determinism: Falsetorchdynamo: Noneray_scope: lastddp_timeout: 1800torch_compile: Falsetorch_compile_backend: Nonetorch_compile_mode: Noneinclude_tokens_per_second: Falseinclude_num_input_tokens_seen: noneftune_noise_alpha: Noneoptim_target_modules: Nonebatch_eval_metrics: Falseeval_on_start: Falseuse_liger_kernel: Falseliger_kernel_config: Noneeval_use_gather_object: Falseaverage_tokens_across_devices: Trueprompts: Nonebatch_sampler: batch_samplermulti_dataset_batch_sampler: round_robinrouter_mapping: {}learning_rate_mapping: {}| Epoch | Step | Training Loss |
|---|---|---|
| 0.1478 | 500 | 5.0049 |
| 0.2956 | 1000 | 4.9978 |
| 0.4434 | 1500 | 4.9986 |
| 0.5912 | 2000 | 4.9993 |
| 0.7390 | 2500 | 5.0005 |
| 0.8868 | 3000 | 4.999 |
| 1.0346 | 3500 | 5.0014 |
| 1.1824 | 4000 | 4.9977 |
| 1.3302 | 4500 | 4.9989 |
| 1.4780 | 5000 | 5.003 |
| 1.6258 | 5500 | 5.0022 |
| 1.7736 | 6000 | 4.9975 |
| 1.9214 | 6500 | 4.9974 |
| 2.0692 | 7000 | 4.9986 |
| 2.2170 | 7500 | 5.0003 |
| 2.3648 | 8000 | 4.9994 |
| 2.5126 | 8500 | 4.9971 |
| 2.6604 | 9000 | 5.0037 |
| 2.8082 | 9500 | 4.997 |
| 2.9560 | 10000 | 4.9997 |
@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",
}
@misc{hermans2017defense,
title={In Defense of the Triplet Loss for Person Re-Identification},
author={Alexander Hermans and Lucas Beyer and Bastian Leibe},
year={2017},
eprint={1703.07737},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
Base model
dangvantuan/vietnamese-document-embedding