metadata

tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:72
  - loss:BatchAllTripletLoss
base_model: cl-nagoya/sup-simcse-ja-base
widget:
  - source_sentence: 打放し型枠（B種）
    sentences:
      - 埋込み（B種）（手間）
      - 埋込み（C種）（手間）
      - 盛土A種
  - source_sentence: 埋込み[B種]
    sentences:
      - 打放し型枠（A種）
      - 盛土（C種）（手間）
      - 埋戻し[C種]
  - source_sentence: 盛土[C種]
    sentences:
      - 埋込み[C種]
      - 盛土（A種）
      - 盛土[A種]
  - source_sentence: 埋戻し[A種]
    sentences:
      - 打放し型枠C種
      - 打放し型枠（C種）（損料・手間）
      - 盛土[B種]
  - source_sentence: 埋込み（B種）（損料・手間）
    sentences:
      - 埋戻し（A種）（損料）
      - 埋戻し（C種）（損料・手間）
      - 埋戻し（B種）（手間）
pipeline_tag: sentence-similarity
library_name: sentence-transformers

SentenceTransformer based on cl-nagoya/sup-simcse-ja-base

This is a sentence-transformers model finetuned from cl-nagoya/sup-simcse-ja-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: cl-nagoya/sup-simcse-ja-base
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-for-standard-name-v0_9_11")
# Run inference
sentences = [
    '埋込み（B種）（損料・手間）',
    '埋戻し（A種）（損料）',
    '埋戻し（B種）（手間）',
]
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: 72 training samples
Columns: sentence and label

Approximate statistics based on the first 72 samples:

	sentence	label
type	string	int
details	min: 11 tokens mean: 16.21 tokens max: 27 tokens	0: ~0.50% 1: ~0.50% 2: ~0.50% 3: ~0.50% 4: ~0.50% 5: ~0.50% 6: ~0.50% 7: ~0.50% 8: ~0.50% 9: ~0.50% 10: ~0.50% 11: ~0.50% 12: ~0.50% 13: ~0.50% 14: ~0.50% 15: ~0.50% 16: ~0.50% 17: ~0.50% 18: ~0.50% 19: ~0.50% 20: ~0.50% 21: ~0.50% 22: ~0.50% 23: ~0.50% 24: ~0.50% 25: ~0.50% 26: ~0.50% 27: ~0.50% 28: ~0.50% 29: ~0.50% 30: ~0.50% 31: ~0.50% 32: ~0.50% 33: ~0.50% 34: ~0.50% 35: ~0.50% 36: ~0.50% 37: ~0.50% 38: ~0.50% 39: ~0.50% 40: ~0.50% 41: ~0.50% 42: ~0.50% 43: ~0.50% 44: ~0.60% 45: ~0.70% 46: ~0.50% 47: ~0.50% 48: ~0.50% 49: ~0.50% 50: ~0.50% 51: ~0.50% 52: ~0.50% 53: ~0.50% 54: ~0.50% 55: ~0.50% 56: ~0.50% 57: ~0.80% 58: ~0.50% 59: ~0.50% 60: ~0.50% 61: ~0.50% 62: ~0.50% 63: ~0.50% 64: ~0.50% 65: ~0.50% 66: ~0.50% 67: ~0.50% 68: ~0.50% 69: ~0.50% 70: ~0.50% 71: ~0.50% 72: ~0.50% 73: ~0.50% 74: ~0.50% 75: ~0.50% 76: ~0.50% 77: ~0.50% 78: ~0.50% 79: ~0.50% 80: ~0.50% 81: ~0.50% 82: ~0.50% 83: ~0.50% 84: ~0.50% 85: ~0.50% 86: ~0.50% 87: ~0.50% 88: ~0.60% 89: ~0.50% 90: ~0.50% 91: ~0.50% 92: ~0.50% 93: ~0.50% 94: ~0.50% 95: ~1.20% 96: ~1.70% 97: ~3.90% 98: ~0.50% 99: ~0.50% 100: ~0.50% 101: ~0.60% 102: ~0.50% 103: ~0.50% 104: ~0.50% 105: ~0.50% 106: ~0.50% 107: ~1.20% 108: ~0.50% 109: ~0.50% 110: ~0.50% 111: ~0.50% 112: ~0.50% 113: ~0.50% 114: ~0.50% 115: ~0.50% 116: ~0.50% 117: ~0.50% 118: ~0.50% 119: ~0.50% 120: ~0.50% 121: ~0.50% 122: ~0.50% 123: ~0.50% 124: ~0.50% 125: ~0.50% 126: ~0.50% 127: ~0.50% 128: ~0.50% 129: ~0.50% 130: ~0.50% 131: ~0.50% 132: ~0.50% 133: ~0.50% 134: ~0.50% 135: ~0.50% 136: ~0.50% 137: ~0.50% 138: ~0.50% 139: ~0.50% 140: ~0.50% 141: ~0.50% 142: ~0.50% 143: ~0.50% 144: ~0.50% 145: ~0.50% 146: ~0.70% 147: ~0.50% 148: ~3.10% 149: ~0.50% 150: ~2.30% 151: ~0.50% 152: ~0.50% 153: ~0.50% 154: ~0.50% 155: ~0.50% 156: ~0.50% 157: ~0.50% 158: ~0.50% 159: ~0.50% 160: ~0.50% 161: ~0.50% 162: ~0.50% 163: ~0.50% 164: ~0.50% 165: ~0.50% 166: ~0.50% 167: ~0.50% 168: ~0.50% 169: ~0.50% 170: ~0.50% 171: ~0.50% 172: ~0.50% 173: ~0.50% 174: ~0.50% 175: ~0.50% 176: ~0.50% 177: ~0.10%

Samples:

sentence	label
`科目：コンクリート。名称：免震基礎天端ｸﾞﾗｳﾄ注入。`	`0`
`科目：コンクリート。名称：免震基礎天端ｸﾞﾗｳﾄ注入。`	`0`
`科目：コンクリート。名称：免震基礎天端ｸﾞﾗｳﾄ注入。`	`0`

Loss: BatchAllTripletLoss

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 512
per_device_eval_batch_size: 512
learning_rate: 1e-05
weight_decay: 0.01
num_train_epochs: 250
warmup_ratio: 0.1
fp16: True
batch_sampler: group_by_label

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: no
prediction_loss_only: True
per_device_train_batch_size: 512
per_device_eval_batch_size: 512
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: 250
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: 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}
tp_size: 0
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: group_by_label
multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss
10.0	10	1.6508
20.0	20	1.2554
30.0	30	0.8495
40.0	40	0.7182
50.0	50	0.6614
60.0	60	0.575
70.0	70	0.5027
80.0	80	0.32
90.0	90	0.1543
100.0	100	0.0102
110.0	110	0.012
120.0	120	0.1164
130.0	130	0.0
140.0	140	0.0
150.0	150	0.0
160.0	160	0.0157
170.0	170	0.0794
180.0	180	0.0
190.0	190	0.0
200.0	200	0.0141
210.0	210	0.0
220.0	220	0.0
230.0	230	0.1115
240.0	240	0.0
250.0	250	0.0
260.0	260	0.0
270.0	270	0.0
280.0	280	0.0
290.0	290	0.0
300.0	300	0.0
310.0	310	0.0
320.0	320	0.0
330.0	330	0.0
340.0	340	0.0
350.0	350	0.0
360.0	360	0.0197
370.0	370	0.0649
380.0	380	0.0
390.0	390	0.0
400.0	400	0.0
410.0	410	0.0
420.0	420	0.0
430.0	430	0.0
440.0	440	0.0
450.0	450	0.0
460.0	460	0.0
470.0	470	0.0
480.0	480	0.0
490.0	490	0.0
500.0	500	0.0
3.1842	100	0.6748
6.3684	200	0.5883
9.5526	300	0.5815
12.7368	400	0.5338
16.1053	500	0.5498
19.2895	600	0.5359
22.4737	700	0.5359
25.6579	800	0.4893
29.0263	900	0.4665
32.2105	1000	0.4205
35.3947	1100	0.4383
38.5789	1200	0.4552
41.7632	1300	0.4003
45.1316	1400	0.3816
48.3158	1500	0.3744
51.5	1600	0.3504
54.6842	1700	0.359
58.0526	1800	0.3019
61.2368	1900	0.3109
64.4211	2000	0.3151
67.6053	2100	0.3292
70.7895	2200	0.2813
74.1579	2300	0.2697
77.3421	2400	0.1975
80.5263	2500	0.2492
83.7105	2600	0.2608
87.0789	2700	0.2401
90.2632	2800	0.2265
93.4474	2900	0.2032
96.6316	3000	0.2368
99.8158	3100	0.2066
103.1842	3200	0.1558
106.3684	3300	0.2029
109.5526	3400	0.244
112.7368	3500	0.1894
116.1053	3600	0.193
119.2895	3700	0.1769
122.4737	3800	0.1821
125.6579	3900	0.0912
129.0263	4000	0.1834
132.2105	4100	0.1391
135.3947	4200	0.1718
138.5789	4300	0.1585
141.7632	4400	0.1829
145.1316	4500	0.1246
148.3158	4600	0.1327
151.5	4700	0.1396
154.6842	4800	0.1028
158.0526	4900	0.0907
161.2368	5000	0.1179
164.4211	5100	0.1496
167.6053	5200	0.1156
170.7895	5300	0.1148
174.1579	5400	0.1275
177.3421	5500	0.1354
180.5263	5600	0.1334
183.7105	5700	0.0874
187.0789	5800	0.0922
190.2632	5900	0.1109
193.4474	6000	0.0708
196.6316	6100	0.0943
199.8158	6200	0.1164
203.1842	6300	0.0785
206.3684	6400	0.0853
209.5526	6500	0.0674
212.7368	6600	0.1009
216.1053	6700	0.0846
219.2895	6800	0.078
222.4737	6900	0.0958
225.6579	7000	0.0811
229.0263	7100	0.0452
232.2105	7200	0.0705
235.3947	7300	0.0664
238.5789	7400	0.0501
241.7632	7500	0.0696
245.1316	7600	0.0736
248.3158	7700	0.08

Framework Versions

Python: 3.11.11
Sentence Transformers: 3.4.1
Transformers: 4.50.2
PyTorch: 2.6.0+cu124
Accelerate: 1.5.2
Datasets: 3.5.0
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",
}

BatchAllTripletLoss

@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}
}