metadata

base_model: colorfulscoop/sbert-base-ja
library_name: sentence-transformers
metrics:
  - cosine_accuracy
  - cosine_accuracy_threshold
  - cosine_f1
  - cosine_f1_threshold
  - cosine_precision
  - cosine_recall
  - cosine_ap
  - dot_accuracy
  - dot_accuracy_threshold
  - dot_f1
  - dot_f1_threshold
  - dot_precision
  - dot_recall
  - dot_ap
  - manhattan_accuracy
  - manhattan_accuracy_threshold
  - manhattan_f1
  - manhattan_f1_threshold
  - manhattan_precision
  - manhattan_recall
  - manhattan_ap
  - euclidean_accuracy
  - euclidean_accuracy_threshold
  - euclidean_f1
  - euclidean_f1_threshold
  - euclidean_precision
  - euclidean_recall
  - euclidean_ap
  - max_accuracy
  - max_accuracy_threshold
  - max_f1
  - max_f1_threshold
  - max_precision
  - max_recall
  - max_ap
pipeline_tag: sentence-similarity
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:5330
  - loss:CoSENTLoss
widget:
  - source_sentence: 建物 に は 、 フェンス の 上 に さまざまな 植物 が 生えて いる 複数 の バルコニー が あり ます 。
    sentences:
      - 建物 の 住人 は とても 良い 庭 師 です 。
      - 男 は 木 を 切る 。
      - 彼 の 家 で フラフープ と 遊ぶ 子供
  - source_sentence: 自転車 の 横 に しゃがむ 男 。
    sentences:
      - ホッケープレイミザー 彼 の ショット
      - フットボール の 試合 を 開始 する 準備 が でき ました
      - 男 が 働いて い ます 。
  - source_sentence: 階段 を 降りて いく 男 。
    sentences:
      - 彼 は どこ か に 行き ます 。
      - 野球 選手 が ボール を 打つ
      - 二 人 の 男 が 雪かき を して い ます 。
  - source_sentence: 青い バケツ を 持つ 少女 は 、 桟橋 の 下 で 跳ね ます 。
    sentences:
      - 子供 は 帽子 を かぶって い ました 。
      - 白 と 黒 の ３ 人 の 女の子 が 外 を 散歩 し ます 。
      - 女の子 が 青い バケツ を 運んで い ます 。
  - source_sentence: 浜辺 の 砂 を 掘る 男
    sentences:
      - 男 が ビーチ に い ます 。
      - 数 人 の 若い 女性 が プレー する 準備 を して い ます 。
      - オートバイ は カー ショー でした 。
model-index:
  - name: SentenceTransformer based on colorfulscoop/sbert-base-ja
    results:
      - task:
          type: binary-classification
          name: Binary Classification
        dataset:
          name: custom arc semantics data jp
          type: custom-arc-semantics-data-jp
        metrics:
          - type: cosine_accuracy
            value: 0.724202626641651
            name: Cosine Accuracy
          - type: cosine_accuracy_threshold
            value: 0.949911892414093
            name: Cosine Accuracy Threshold
          - type: cosine_f1
            value: 0.8227550540667607
            name: Cosine F1
          - type: cosine_f1_threshold
            value: 0.9255338907241821
            name: Cosine F1 Threshold
          - type: cosine_precision
            value: 0.7157464212678937
            name: Cosine Precision
          - type: cosine_recall
            value: 0.9673852957435047
            name: Cosine Recall
          - type: cosine_ap
            value: 0.7633272592963735
            name: Cosine Ap
          - type: dot_accuracy
            value: 0.726454033771107
            name: Dot Accuracy
          - type: dot_accuracy_threshold
            value: 626.92529296875
            name: Dot Accuracy Threshold
          - type: dot_f1
            value: 0.8233872916163325
            name: Dot F1
          - type: dot_f1_threshold
            value: 612.754638671875
            name: Dot F1 Threshold
          - type: dot_precision
            value: 0.7313304721030043
            name: Dot Precision
          - type: dot_recall
            value: 0.9419568822553898
            name: Dot Recall
          - type: dot_ap
            value: 0.7865839551255255
            name: Dot Ap
          - type: manhattan_accuracy
            value: 0.724953095684803
            name: Manhattan Accuracy
          - type: manhattan_accuracy_threshold
            value: 180.30792236328125
            name: Manhattan Accuracy Threshold
          - type: manhattan_f1
            value: 0.8225806451612903
            name: Manhattan F1
          - type: manhattan_f1_threshold
            value: 244.3115997314453
            name: Manhattan F1 Threshold
          - type: manhattan_precision
            value: 0.705254839984196
            name: Manhattan Precision
          - type: manhattan_recall
            value: 0.9867330016583747
            name: Manhattan Recall
          - type: manhattan_ap
            value: 0.7637811425109782
            name: Manhattan Ap
          - type: euclidean_accuracy
            value: 0.7238273921200751
            name: Euclidean Accuracy
          - type: euclidean_accuracy_threshold
            value: 8.075063705444336
            name: Euclidean Accuracy Threshold
          - type: euclidean_f1
            value: 0.8225616921269095
            name: Euclidean F1
          - type: euclidean_f1_threshold
            value: 9.857145309448242
            name: Euclidean F1 Threshold
          - type: euclidean_precision
            value: 0.7154538021259199
            name: Euclidean Precision
          - type: euclidean_recall
            value: 0.9673852957435047
            name: Euclidean Recall
          - type: euclidean_ap
            value: 0.7631772892743254
            name: Euclidean Ap
          - type: max_accuracy
            value: 0.726454033771107
            name: Max Accuracy
          - type: max_accuracy_threshold
            value: 626.92529296875
            name: Max Accuracy Threshold
          - type: max_f1
            value: 0.8233872916163325
            name: Max F1
          - type: max_f1_threshold
            value: 612.754638671875
            name: Max F1 Threshold
          - type: max_precision
            value: 0.7313304721030043
            name: Max Precision
          - type: max_recall
            value: 0.9867330016583747
            name: Max Recall
          - type: max_ap
            value: 0.7865839551255255
            name: Max Ap

SentenceTransformer based on colorfulscoop/sbert-base-ja

This is a sentence-transformers model finetuned from colorfulscoop/sbert-base-ja on the csv dataset. 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: colorfulscoop/sbert-base-ja
Maximum Sequence Length: 512 tokens
Output Dimensionality: 768 tokens
Similarity Function: Cosine Similarity
Training Dataset:
- csv

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': 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 = [
    '浜辺 の 砂 を 掘る 男',
    '男 が ビーチ に い ます 。',
    'オートバイ は カー ショー でした 。',
]
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]

Evaluation

Metrics

Binary Classification

Dataset: custom-arc-semantics-data-jp
Evaluated with BinaryClassificationEvaluator

Metric	Value
cosine_accuracy	0.7242
cosine_accuracy_threshold	0.9499
cosine_f1	0.8228
cosine_f1_threshold	0.9255
cosine_precision	0.7157
cosine_recall	0.9674
cosine_ap	0.7633
dot_accuracy	0.7265
dot_accuracy_threshold	626.9253
dot_f1	0.8234
dot_f1_threshold	612.7546
dot_precision	0.7313
dot_recall	0.942
dot_ap	0.7866
manhattan_accuracy	0.725
manhattan_accuracy_threshold	180.3079
manhattan_f1	0.8226
manhattan_f1_threshold	244.3116
manhattan_precision	0.7053
manhattan_recall	0.9867
manhattan_ap	0.7638
euclidean_accuracy	0.7238
euclidean_accuracy_threshold	8.0751
euclidean_f1	0.8226
euclidean_f1_threshold	9.8571
euclidean_precision	0.7155
euclidean_recall	0.9674
euclidean_ap	0.7632
max_accuracy	0.7265
max_accuracy_threshold	626.9253
max_f1	0.8234
max_f1_threshold	612.7546
max_precision	0.7313
max_recall	0.9867
max_ap	0.7866

Training Details

Training Dataset

csv

Dataset: csv
Size: 5,330 training samples
Columns: text1, text2, and label
Approximate statistics based on the first 1000 samples:
text1 text2 label
type string string int
details
min: 7 tokens
mean: 36.64 tokens
max: 103 tokens

min: 8 tokens
mean: 22.81 tokens
max: 77 tokens

0: ~33.30%
1: ~66.70%

	text1	text2	label
type	string	string	int
details	min: 7 tokens mean: 36.64 tokens max: 103 tokens	min: 8 tokens mean: 22.81 tokens max: 77 tokens	0: ~33.30% 1: ~66.70%

Samples:

text1	text2	label
`草の山で寝ている男。`	`男は目を覚ましている。`	`1`
`セーターとジーンズを着ている女性が屋外公園で２人の小さな子供とタイヤスイングで遊んでいます。`	`母親が２人の子供と遊んでいます。`	`1`
`ジョガーは、伸ばした木の枝の下を通ります。`	`ジョガーは別のツリーブランチを回っています。`	`1`

Loss: CoSENTLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "pairwise_cos_sim"
}

Evaluation Dataset

csv

Dataset: csv
Size: 5,330 evaluation samples
Columns: text1, text2, and label
Approximate statistics based on the first 1000 samples:
text1 text2 label
type string string int
details
min: 8 tokens
mean: 35.92 tokens
max: 177 tokens

min: 4 tokens
mean: 22.45 tokens
max: 68 tokens

0: ~33.30%
1: ~66.70%

	text1	text2	label
type	string	string	int
details	min: 8 tokens mean: 35.92 tokens max: 177 tokens	min: 4 tokens mean: 22.45 tokens max: 68 tokens	0: ~33.30% 1: ~66.70%

Samples:

text1	text2	label
`空中でスタントを行うスノーボーダー。`	`危険なスタントを行うスノーボーダー`	`1`
`高級レストランでタイル張りのレストランカレンダーの背後で２人のシェフが語り合います。`	`レストランで食事をする２人のシェフ。`	`1`
`年配の男性が水上で手ｒｏｗぎボートに立っています。`	`人は橋から飛び降りています`	`1`

Loss: CoSENTLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "pairwise_cos_sim"
}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: epoch
learning_rate: 2e-05
num_train_epochs: 1
warmup_ratio: 0.4
fp16: 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: 8
per_device_eval_batch_size: 8
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: 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: 1
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: {}
warmup_ratio: 0.4
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: False
hub_always_push: False
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
include_inputs_for_metrics: False
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
eval_use_gather_object: False
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	loss	custom-arc-semantics-data-jp_max_ap
0	0	-	-	0.5488
1.0	334	3.4293	2.3784	0.7866

Framework Versions

Python: 3.10.14
Sentence Transformers: 3.1.0
Transformers: 4.44.2
PyTorch: 2.4.1+cu121
Accelerate: 0.34.2
Datasets: 2.20.0
Tokenizers: 0.19.1

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",
}

CoSENTLoss

@online{kexuefm-8847,
    title={CoSENT: A more efficient sentence vector scheme than Sentence-BERT},
    author={Su Jianlin},
    year={2022},
    month={Jan},
    url={https://kexue.fm/archives/8847},
}