b00l26/vietnamese_embedding_finetune_news

SentenceTransformer based on dangvantuan/vietnamese-embedding

This is a sentence-transformers model finetuned from dangvantuan/vietnamese-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.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: dangvantuan/vietnamese-embedding
• 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': 'RobertaModel'})
  (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

# Download from the 🤗 Hub
model = SentenceTransformer("sentence_transformers_model_id")
# Run inference
sentences = [
    'Ông trùm nắm chìa_khoá 5.000 tỷ USD : Nhân_vật 2025 ? . ( Dân_trí ) - Nhận cuộc_gọi Trump nửa_đêm nắm đế_chế 5.000 tỷ USD , Jensen_Huang Nhân_vật 2025 . Bí_mật ẩn áo da  quyền_lực  canh_bạc AI tất ?',
    'Đưa_Nvidia chạm mốc 4.200 tỷ USD , CEO Jensen_Huang “ cày ” cỡ ? . ( Dân_trí ) - Để Nvidia thành công_ty đắt_giá thế_giới , Jensen_Huang đánh_đổi đời_sống thường_nhật : Không phim_ảnh , nghỉ , bộ_não “ tắt ” .',
    'Nợ 4,4 tỷ đồng , cụ U90 Trung_Quốc ròng_rã may áo suốt 10 . ( Dân_trí ) - Cụ Chen_Jinying cảm_phục ý_chí nghị_lực ròng_rã áo nợ suốt 10 .',
]
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, 0.6855, 0.1530],
#         [0.6855, 1.0000, 0.1226],
#         [0.1530, 0.1226, 1.0000]])

Evaluation

Metrics

Information Retrieval

• Dataset: ir_val
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.0005
cosine_accuracy@3	0.4075
cosine_accuracy@5	0.558
cosine_accuracy@10	0.7111
cosine_precision@1	0.0005
cosine_precision@3	0.1637
cosine_precision@5	0.1598
cosine_precision@10	0.1217
cosine_recall@1	0.0001
cosine_recall@3	0.2533
cosine_recall@5	0.3939
cosine_recall@10	0.5643
cosine_ndcg@10	0.3111
cosine_mrr@10	0.2362
cosine_map@10	0.1989

Training Details

Training Dataset

Unnamed Dataset

• Size: 39,595 training samples
• Columns: sentence_0 and sentence_1
• Approximate statistics based on the first 1000 samples:

  	sentence_0	sentence_1
  type	string	string
  details	min: 16 tokens mean: 33.74 tokens max: 87 tokens	min: 14 tokens mean: 34.16 tokens max: 86 tokens

• Samples:

  sentence_0	sentence_1
  Đạo_diễn , NSND Thanh_Vân cay_đắng Hãng phim_truyện Việt_Nam . ( Dân_trí ) - Trước thảm_cảnh Hãng phim_truyện Việt_Nam , đạo_diễn Thanh_Vân PV Dân_trí : Tôi cay_đắng 30 tận_tâm cống_hiến , Hãng , nằm viện 100% viện_phí .	Đạo_diễn Luk_Vân áp_lực phim Công nữ Ngọc_Hoa . ( Dân_trí ) - Ngày 22/11 , đạo_diễn Luk_Vân công_bố dự_án Công nữ Ngọc_Hoa , phim_điện_ảnh hợp_tác Việt - Nhật . Ý_tưởng phim tình đẹp Công nữ Ngọc_Hoa thương_nhân Nhật_Bản Araki_Sotaro .
  Bộ_trưởng Tài_chính : Việt_Nam cố_gắng đáp_ứng tiêu_chí nâng hạng FTSE . Việt_Nam nỗ_lực đáp_ứng tiêu_chí nâng hạng FTSE thông_qua cải_cách thuận_lợi dòng vốn đầu_tư nước_ngoài thị_trường , Bộ_trưởng Nguyễn_Văn_Thắng .	Chứng_khoán . VN-Index giằng_co quyết_liệt giao_dịch đột_ngột rớt 10 phiên khớp lệnh xác_định giá đóng_cửa .
  Nhóm BTOB tái_ngộ khán_giả Việt . Nhóm nhạc Hàn_Quốc BTOB trở_lại sân_khấu TP HCM bảy , khuấy_động không_khí loạt hit , tối 31/10 .	Hanbin - chàng trai Việt toả thần_tượng Kpop . Hanbin ( Ngô_Ngọc_Hưng ) - 26 , quê Yên_Bái - fan đông_đảo nhạc Hàn_TEMPEST giọng hát , vũ_đạo nổi_bật .

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 32
• num_train_epochs: 5
• fp16: True
• multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 32
• 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: 5
• 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
• bf16: False
• fp16: True
• 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}
• parallelism_config: None
• deepspeed: None
• label_smoothing_factor: 0.0
• optim: adamw_torch_fused
• optim_args: None
• adafactor: False
• 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
• 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: 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
• include_tokens_per_second: False
• 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
• prompts: None
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: round_robin
• router_mapping: {}
• learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss	ir_val_cosine_ndcg@10
0.0808	50	-	0.1715
0.1616	100	-	0.1863
0.2423	150	-	0.1975
0.3231	200	-	0.2170
0.4039	250	-	0.2236
0.4847	300	-	0.2303
0.5654	350	-	0.2344
0.6462	400	-	0.2353
0.7270	450	-	0.2371
0.8078	500	0.8206	0.2399
0.8885	550	-	0.2423
0.9693	600	-	0.2440
1.0	619	-	0.2469
1.0501	650	-	0.2485
1.1309	700	-	0.2504
1.2116	750	-	0.2542
1.2924	800	-	0.2582
1.3732	850	-	0.2538
1.4540	900	-	0.2586
1.5347	950	-	0.2612
1.6155	1000	0.4315	0.2613
1.6963	1050	-	0.2608
1.7771	1100	-	0.2625
1.8578	1150	-	0.2658
1.9386	1200	-	0.2674
2.0	1238	-	0.2675
2.0194	1250	-	0.2695
2.1002	1300	-	0.2730
2.1809	1350	-	0.2745
2.2617	1400	-	0.2772
2.3425	1450	-	0.2800
2.4233	1500	0.3189	0.2783
2.5040	1550	-	0.2793
2.5848	1600	-	0.2810
2.6656	1650	-	0.2804
2.7464	1700	-	0.2821
2.8271	1750	-	0.2850
2.9079	1800	-	0.2846
2.9887	1850	-	0.2857
3.0	1857	-	0.2850
3.0695	1900	-	0.2874
3.1502	1950	-	0.2869
3.2310	2000	0.2421	0.2878
3.3118	2050	-	0.2894
3.3926	2100	-	0.2932
3.4733	2150	-	0.2959
3.5541	2200	-	0.2954
3.6349	2250	-	0.2951
3.7157	2300	-	0.2986
3.7964	2350	-	0.3013
3.8772	2400	-	0.2980
3.9580	2450	-	0.2992
4.0	2476	-	0.3006
4.0388	2500	0.2048	0.3005
4.1195	2550	-	0.3019
4.2003	2600	-	0.3037
4.2811	2650	-	0.3038
4.3619	2700	-	0.3045
4.4426	2750	-	0.3068
4.5234	2800	-	0.3087
4.6042	2850	-	0.3074
4.6850	2900	-	0.3082
4.7658	2950	-	0.3078
4.8465	3000	0.1815	0.3086
4.9273	3050	-	0.3076
5.0	3095	-	0.3111

Framework Versions

• Python: 3.12.12
• Sentence Transformers: 5.1.1
• Transformers: 4.57.1
• PyTorch: 2.8.0+cu126
• Accelerate: 1.11.0
• Datasets: 4.4.1
• Tokenizers: 0.22.1

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}
}