Detomo/cl-nagoya-sup-simcse-ja-nss-v1_0_8

SentenceTransformer

This is a sentence-transformers model trained. 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
• Maximum Sequence Length: 512 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': 512, 'do_lower_case': False}) with Transformer model: BertModel 
  (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})
)

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("Detomo/cl-nagoya-sup-simcse-ja-nss-v1_0_8")
# Run inference
sentences = [
    '科目：コンクリート。名称：基礎部ｺﾝｸﾘｰﾄ打設手間。',
    '科目：コンクリート。名称：底盤ｺﾝｸﾘｰﾄ打設手間。',
    '科目：タイル。名称：外壁ｶﾞﾗｽﾓｻﾞｲｸﾀｲﾙ張り。',
]
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: 327,543 training samples
• Columns: sentence1, sentence2, and label
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2	label
  type	string	string	int
  details	min: 11 tokens mean: 13.78 tokens max: 19 tokens	min: 11 tokens mean: 14.8 tokens max: 23 tokens	0: ~74.10% 1: ~2.60% 2: ~23.30%

• Samples:

  sentence1	sentence2	label
  科目：コンクリート。名称：免震基礎天端ｸﾞﾗｳﾄ注入。	科目：コンクリート。名称：免震BPL下部充填ｺﾝｸﾘｰﾄ打設手間。	0
  科目：コンクリート。名称：免震基礎天端ｸﾞﾗｳﾄ注入。	科目：コンクリート。名称：免震下部（外周基礎梁）ｺﾝｸﾘｰﾄ打設手間。	0
  科目：コンクリート。名称：免震基礎天端ｸﾞﾗｳﾄ注入。	科目：コンクリート。名称：免震下部ｺﾝｸﾘｰﾄ打設手間。	0

• Loss: sentence_transformer_lib.categorical_constrastive_loss.CategoricalContrastiveLoss

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 256
• per_device_eval_batch_size: 256
• learning_rate: 1e-05
• weight_decay: 0.01
• num_train_epochs: 4
• warmup_ratio: 0.2
• fp16: True

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: no
• prediction_loss_only: True
• per_device_train_batch_size: 256
• per_device_eval_batch_size: 256
• 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: 1e-05
• weight_decay: 0.01
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1.0
• num_train_epochs: 4
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.2
• 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: 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}
• 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
• 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: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss
0.0391	50	0.0432
0.0781	100	0.0449
0.1172	150	0.0429
0.1562	200	0.0397
0.1953	250	0.0395
0.2344	300	0.0312
0.2734	350	0.0347
0.3125	400	0.0303
0.3516	450	0.0298
0.3906	500	0.0321
0.4297	550	0.0266
0.4688	600	0.0254
0.5078	650	0.0267
0.5469	700	0.0244
0.5859	750	0.0238
0.625	800	0.0229
0.6641	850	0.023
0.7031	900	0.0189
0.7422	950	0.0207
0.7812	1000	0.0201
0.8203	1050	0.0188
0.8594	1100	0.0153
0.8984	1150	0.0168
0.9375	1200	0.014
0.9766	1250	0.0155
1.0156	1300	0.0141
1.0547	1350	0.0139
1.0938	1400	0.0121
1.1328	1450	0.0121
1.1719	1500	0.0109
1.2109	1550	0.0116
1.25	1600	0.0119
1.2891	1650	0.0102
1.3281	1700	0.0095
1.3672	1750	0.0089
1.4062	1800	0.0109
1.4453	1850	0.0094
1.4844	1900	0.0094
1.5234	1950	0.0089
1.5625	2000	0.0088
1.6016	2050	0.0081
1.6406	2100	0.0082
1.6797	2150	0.0072
1.7188	2200	0.0075
1.7578	2250	0.0078
1.7969	2300	0.0081
1.8359	2350	0.0079
1.875	2400	0.008
1.9141	2450	0.0079
1.9531	2500	0.0071
1.9922	2550	0.0089
2.0312	2600	0.0063
2.0703	2650	0.0055
2.1094	2700	0.0053
2.1484	2750	0.0053
2.1875	2800	0.0054
2.2266	2850	0.0046
2.2656	2900	0.005
2.3047	2950	0.0053
2.3438	3000	0.0047
2.3828	3050	0.0052
2.4219	3100	0.0049
2.4609	3150	0.0055
2.5	3200	0.0047
2.5391	3250	0.0048
2.5781	3300	0.0046
2.6172	3350	0.0049
2.6562	3400	0.0049
2.6953	3450	0.0051
2.7344	3500	0.0045
2.7734	3550	0.0044
2.8125	3600	0.0049
2.8516	3650	0.0048
2.8906	3700	0.0047
2.9297	3750	0.0044
2.9688	3800	0.0041
3.0078	3850	0.0039
3.0469	3900	0.0038
3.0859	3950	0.0033
3.125	4000	0.0037
3.1641	4050	0.0036
3.2031	4100	0.004
3.2422	4150	0.0036
3.2812	4200	0.0038
3.3203	4250	0.004
3.3594	4300	0.004
3.3984	4350	0.0039
3.4375	4400	0.0031
3.4766	4450	0.0031
3.5156	4500	0.0038
3.5547	4550	0.0031
3.5938	4600	0.0029
3.6328	4650	0.0031
3.6719	4700	0.003
3.7109	4750	0.0036
3.75	4800	0.0035
3.7891	4850	0.0029
3.8281	4900	0.0033
3.8672	4950	0.0031
3.9062	5000	0.0036
3.9453	5050	0.0037
3.9844	5100	0.0031

Framework Versions

• Python: 3.11.13
• Sentence Transformers: 4.1.0
• Transformers: 4.52.4
• PyTorch: 2.6.0+cu124
• Accelerate: 1.7.0
• Datasets: 2.14.4
• Tokenizers: 0.21.1

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",
}