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:53
  - loss:CosineSimilarityLoss
widget:
  - source_sentence: 黒い タイル の 本当に すてきな カウンター の 前 と 後ろ で 働く 人々 。
    sentences:
      - 男性 は バレエ に 参加 して い ます 。
      - 岩 の 上 に 座って いる 二 人
      - 人々 は 宝石 店 で 働いて い ます 。
  - source_sentence: 少年 は 木 の 切り株 に 座って い ます 。
    sentences:
      - ストリート ワーカー は 保護 具 を 着用 して い ませ ん 。
      - 芝生 の エリア で 数 匹 の 犬 が 交流 し ます 。
      - 木 を 切り 倒した 後 、 木 の 切り株 に 座って いる 少年 。
  - source_sentence: 岩 の 多い 景色 を 見て 二 人
    sentences:
      - 何 か を 見て いる 二 人 が い ます 。
      - 誰 か が 肖像 画 を 描いて い ます 。
      - バイカー は 足 を 使って 自転車 を さらに 進め ます 。
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: 0.6363636363636364
            name: Cosine Accuracy
          - type: cosine_accuracy_threshold
            value: 0.3379952907562256
            name: Cosine Accuracy Threshold
          - type: cosine_f1
            value: 0.7777777777777777
            name: Cosine F1
          - type: cosine_f1_threshold
            value: 0.3379952907562256
            name: Cosine F1 Threshold
          - type: cosine_precision
            value: 0.7
            name: Cosine Precision
          - type: cosine_recall
            value: 0.875
            name: Cosine Recall
          - type: cosine_ap
            value: 0.619629329004329
            name: Cosine Ap
          - type: dot_accuracy
            value: 0.6363636363636364
            name: Dot Accuracy
          - type: dot_accuracy_threshold
            value: 187.5118865966797
            name: Dot Accuracy Threshold
          - type: dot_f1
            value: 0.7777777777777777
            name: Dot F1
          - type: dot_f1_threshold
            value: 187.5118865966797
            name: Dot F1 Threshold
          - type: dot_precision
            value: 0.7
            name: Dot Precision
          - type: dot_recall
            value: 0.875
            name: Dot Recall
          - type: dot_ap
            value: 0.6946293290043289
            name: Dot Ap
          - type: manhattan_accuracy
            value: 0.6363636363636364
            name: Manhattan Accuracy
          - type: manhattan_accuracy_threshold
            value: 598.9317626953125
            name: Manhattan Accuracy Threshold
          - type: manhattan_f1
            value: 0.7777777777777777
            name: Manhattan F1
          - type: manhattan_f1_threshold
            value: 598.9317626953125
            name: Manhattan F1 Threshold
          - type: manhattan_precision
            value: 0.7
            name: Manhattan Precision
          - type: manhattan_recall
            value: 0.875
            name: Manhattan Recall
          - type: manhattan_ap
            value: 0.619629329004329
            name: Manhattan Ap
          - type: euclidean_accuracy
            value: 0.6363636363636364
            name: Euclidean Accuracy
          - type: euclidean_accuracy_threshold
            value: 27.118305206298828
            name: Euclidean Accuracy Threshold
          - type: euclidean_f1
            value: 0.7777777777777777
            name: Euclidean F1
          - type: euclidean_f1_threshold
            value: 27.118305206298828
            name: Euclidean F1 Threshold
          - type: euclidean_precision
            value: 0.7
            name: Euclidean Precision
          - type: euclidean_recall
            value: 0.875
            name: Euclidean Recall
          - type: euclidean_ap
            value: 0.619629329004329
            name: Euclidean Ap
          - type: max_accuracy
            value: 0.6363636363636364
            name: Max Accuracy
          - type: max_accuracy_threshold
            value: 598.9317626953125
            name: Max Accuracy Threshold
          - type: max_f1
            value: 0.7777777777777777
            name: Max F1
          - type: max_f1_threshold
            value: 598.9317626953125
            name: Max F1 Threshold
          - type: max_precision
            value: 0.7
            name: Max Precision
          - type: max_recall
            value: 0.875
            name: Max Recall
          - type: max_ap
            value: 0.6946293290043289
            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 = [
    '岩 の 多い 景色 を 見て 二 人',
    '何 か を 見て いる 二 人 が い ます 。',
    '誰 か が 肖像 画 を 描いて い ます 。',
]
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	0.6364
cosine_accuracy_threshold	0.338
cosine_f1	0.7778
cosine_f1_threshold	0.338
cosine_precision	0.7
cosine_recall	0.875
cosine_ap	0.6196
dot_accuracy	0.6364
dot_accuracy_threshold	187.5119
dot_f1	0.7778
dot_f1_threshold	187.5119
dot_precision	0.7
dot_recall	0.875
dot_ap	0.6946
manhattan_accuracy	0.6364
manhattan_accuracy_threshold	598.9318
manhattan_f1	0.7778
manhattan_f1_threshold	598.9318
manhattan_precision	0.7
manhattan_recall	0.875
manhattan_ap	0.6196
euclidean_accuracy	0.6364
euclidean_accuracy_threshold	27.1183
euclidean_f1	0.7778
euclidean_f1_threshold	27.1183
euclidean_precision	0.7
euclidean_recall	0.875
euclidean_ap	0.6196
max_accuracy	0.6364
max_accuracy_threshold	598.9318
max_f1	0.7778
max_f1_threshold	598.9318
max_precision	0.7
max_recall	0.875
max_ap	0.6946

Training Details

Training Dataset

csv

Dataset: csv
Size: 53 training samples
Columns: text1, text2, and label
Approximate statistics based on the first 53 samples:
text1 text2 label
type string string int
details
min: 14 tokens
mean: 35.36 tokens
max: 79 tokens

min: 11 tokens
mean: 21.33 tokens
max: 38 tokens

0: ~38.10%
1: ~61.90%

	text1	text2	label
type	string	string	int
details	min: 14 tokens mean: 35.36 tokens max: 79 tokens	min: 11 tokens mean: 21.33 tokens max: 38 tokens	0: ~38.10% 1: ~61.90%

Samples:

text1	text2	label
`薄紫色のドレスと明るいホットピンクの靴を着た女性が、水とコーヒーを飲んでテーブルに座っています。`	`ブラインドデートの女性が座って、デートが現れるのを待ちます。`	`1`
`トラックを自転車で走る人々のグループ。`	`自転車の挑戦に勝とうとする人々のグループ。`	`1`
`野球の試合中に基地を走る野球選手のシャープリー。`	`Ｓｈａｒｐｌｅｙはゲームでプレイしています。`	`0`

Loss: CosineSimilarityLoss with these parameters:

{
    "loss_fct": "torch.nn.modules.loss.MSELoss"
}

Evaluation Dataset

csv

Dataset: csv
Size: 53 evaluation samples
Columns: text1, text2, and label
Approximate statistics based on the first 53 samples:
text1 text2 label
type string string int
details
min: 19 tokens
mean: 39.64 tokens
max: 84 tokens

min: 19 tokens
mean: 25.27 tokens
max: 38 tokens

0: ~27.27%
1: ~72.73%

	text1	text2	label
type	string	string	int
details	min: 19 tokens mean: 39.64 tokens max: 84 tokens	min: 19 tokens mean: 25.27 tokens max: 38 tokens	0: ~27.27% 1: ~72.73%

Samples:

text1	text2	label
`岩の多い景色を見て二人`	`何かを見ている二人がいます。`	`0`
`白いヘルメットとオレンジ色のシャツ、ジーンズ、白いトラックとオレンジ色のパイロンの前に反射ジャケットを着た金髪のストリートワーカー。`	`ストリートワーカーは保護具を着用していません。`	`1`
`白い帽子をかぶった女性が、鮮やかな色の岩の風景を描いています。岩層自体が背景に見えます。`	`誰かが肖像画を描いています。`	`1`

Loss: CosineSimilarityLoss with these parameters:

{
    "loss_fct": "torch.nn.modules.loss.MSELoss"
}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: epoch
learning_rate: 2e-05
num_train_epochs: 15
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: 15
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
1.0	6	0.2963	0.3111	0.6821
2.0	12	0.2833	0.3096	0.7238
3.0	18	0.2568	0.3050	0.7238
4.0	24	0.2177	0.2958	0.7238
5.0	30	0.1797	0.2826	0.6946
6.0	36	0.1419	0.2765	0.6509
7.0	42	0.1057	0.2954	0.6509
8.0	48	0.0815	0.3165	0.6509
9.0	54	0.0664	0.3199	0.6509
10.0	60	0.0497	0.3140	0.6509
11.0	66	0.0402	0.3081	0.6321
12.0	72	0.0346	0.3072	0.6946
13.0	78	0.0293	0.3066	0.6946
14.0	84	0.0302	0.3076	0.6946
15.0	90	0.0287	0.3078	0.6946

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",
}