SentenceTransformer based on sergeyzh/rubert-tiny-turbo
This is a sentence-transformers model finetuned from sergeyzh/rubert-tiny-turbo. It maps sentences & paragraphs to a 312-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: sergeyzh/rubert-tiny-turbo
- Maximum Sequence Length: 2048 tokens
- Output Dimensionality: 312 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': 2048, 'do_lower_case': False}) with Transformer model: BertModel
(1): Pooling({'word_embedding_dimension': 312, '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})
(2): Normalize()
)
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("denis-gordeev/reranker_dialog_items_biencoder_rubert-tiny-turbo-5")
# Run inference
sentences = [
'расскажи о камерах смартфонов',
"{'long_web_name': 'Смартфон Honor 200 Lite 8/256GB голубой (5109BFBH)', 'price': 21290.0, 'description': '', 'rating': 4.83, 'review_count': 17}",
"{'long_web_name': 'Накладка силикон для Xiaomi Redmi 5 (оригинальный) прозрачный', 'price': 599.0, 'url': 'https://megamarket.ru/catalog/details/nakladka-silikon-dlya-xiaomi-redmi-5-originalnyy-prozrachnyy-100057155753/', 'image_link': 'https://main-cdn.sbermegamarket.ru/mid9/hlr-system/-24/417/121/310/276/47/100057155753b0.jpg', 'id': '100057155753_102580', 'description': '', 'rating': 0.0, 'review_count': 0}",
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 312]
# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]
Evaluation
Metrics
Binary Classification
- Dataset:
item-classification
- Evaluated with
BinaryClassificationEvaluator
Metric | Value |
---|---|
cosine_accuracy | 0.9843 |
cosine_accuracy_threshold | 0.7253 |
cosine_f1 | 0.9494 |
cosine_f1_threshold | 0.7253 |
cosine_precision | 0.9298 |
cosine_recall | 0.9698 |
cosine_ap | 0.9839 |
Training Details
Training Dataset
Unnamed Dataset
- Size: 48,868 training samples
- Columns:
anchor
,text
, andlabel
- Approximate statistics based on the first 1000 samples:
anchor text label type string string int details - min: 5 tokens
- mean: 17.78 tokens
- max: 118 tokens
- min: 57 tokens
- mean: 318.85 tokens
- max: 1182 tokens
- 0: ~85.50%
- 1: ~14.50%
- Samples:
anchor text label помоги подобрать внешний аккумулятор, чтобы получить сбербонусы
покажи товары{'long_web_name': 'Чехол для Xiaomi Battery Case 10000mAh ver.2 Orange', 'price': 195.0, 'url': 'https://megamarket.ru/catalog/details/xiaomi-battery-case-10000mah-ver2-orange-100043272924/', 'image_link': 'https://main-cdn.sbermegamarket.ru/mid9/hlr-system/202/591/442/682/916/55/100043272924b0.jpg', 'id': '100043272924', 'description': '', 'rating': 0.0, 'review_count': 0}
0
Здравствуйте. Мне нужен недорогой смартфон на Android, чтобы можно было легко звонить и писать сообщения внукам. Можете что-то посоветовать?
{'long_web_name': 'Чистящее средство Topperr 3037', 'price': 417.0, 'url': 'https://megamarket.ru/catalog/details/chistyashee-sredstvo-dlya-kofemashin-topperr-3037-100022709014/', 'image_link': 'https://main-cdn.sbermegamarket.ru/mid9/hlr-system/-16/699/502/081/231/16/100022709014b0.jpg', 'id': '100022709014', 'description': '', 'rating': 4.94, 'review_count': 222}
0
Samsung Galaxy S24
{'long_web_name': 'Поворотное металлическое крепление на руль мотоцикла велосипеда для экшн камеры GoPro', 'price': 950.0, 'url': 'https://megamarket.ru/catalog/details/kreplenie-nobrand-00000659-600016461568/', 'image_link': 'https://main-cdn.sbermegamarket.ru/mid9/hlr-system/-66/144/792/042/153/1/600016461568b0.png', 'id': '600016461568_81689', 'description': '
Это надежный металлический крепеж для рулей и круглых труб небольшого диаметра, до 33мм. Крепление оснащено стандартным U-образным креплением, которое совместимо с экшн камерами GoPro, SjCam, Xiaomi и иных других оснащенных подобным креплением. Есть возможность поворота камеры вокруг своей оси на 360 градусов с фиксацией.
Благодаря резиновым уплотнителям внутри, крепление надежно держится и не провернется вокруг трубы.
Крепление затягивается шестигранником, который идет в комплекте. Также в комплекте идет металлический болт для закрепления экшн камеры в U-образном креплении.
', 'rating': 0.0, 'review_count': ...0
- Loss:
ContrastiveLoss
with these parameters:{ "distance_metric": "SiameseDistanceMetric.COSINE_DISTANCE", "margin": 0.5, "size_average": true }
Evaluation Dataset
Unnamed Dataset
- Size: 6,108 evaluation samples
- Columns:
anchor
,text
, andlabel
- Approximate statistics based on the first 1000 samples:
anchor text label type string string int details - min: 5 tokens
- mean: 20.51 tokens
- max: 1716 tokens
- min: 53 tokens
- mean: 326.79 tokens
- max: 1182 tokens
- 0: ~84.50%
- 1: ~15.50%
- Samples:
anchor text label Привет, помоги подобрать ноутбук, на что обратить внимание?
Диагональ дисплея хочу 15
оеративка от 16гб
ссд хотя бы 0.5 тб
Порекомендуй конкртные товары{'long_web_name': 'Ноутбук Azerty RB-1550 Silver (120-0513)', 'price': 25470.0, 'description': 'Ноутбук Azerty RB-1550 обладает достаточной производительностью для решения учебных задач, таких как работа с документами, просмотр веб-страниц, использование электронных учебников и презентаций. Конечно, ведь именно для этого он и предназначен, являясь представителем серии моделей для учёбы Story. - Корпус ноутбука выполнен из твердого полимерного пластика, придающего легкости и прочности. Этот материал обладает высокой устойчивостью к механическим воздействиям, царапинам и потёртостям, а также снижает вес устройства, что делает его удобным для переноски. - Экран ноутбука имеет размер 15,6 дюйма выполнен по технологии IPS, которая обеспечивает хорошее качество изображения, с высокой контрастностью и широким углом обзора. Разрешение экрана составляет 1920x1080 пикселей, что обеспечивает высокую детальность. Антибликовое покрытие экрана помогает снизить нагрузку на зрение при работе в ярко ос...
0
расскажи как выбрать смартфон игровой
расскажи о разнице между андроидом и айос подробнее
расскажи подробнее об операционной системе{'long_web_name': 'Смартфон Honor Honor 90 12/512GB изумрудный зеленый (5109ATRU)', 'price': 33990.0, 'description': '', 'rating': 4.73, 'review_count': 37}
1
Найди самсунг белого цвета в республике башкортостан. Меня зовут Алексей, кстати
И до 50к с 8 гб оперативы{'long_web_name': 'Защитное стекло Remax Medicine Glass GL-27 3D для iPhone 15, черная рамка 0,3 мм', 'price': 247.0, 'url': 'https://megamarket.ru/catalog/details/zashitnoe-steklo-remax-medicine-glass-gl-27-3d-dlya-iphone-15-chernaya-ramka-03-mm-600013601251/', 'image_link': 'https://main-cdn.sbermegamarket.ru/mid9/hlr-system/811/370/081/107/016/600013601251b0.jpeg', 'id': '600013601251', 'description': 'Защитное стекло для Apple iPhone 15/ Айфон 15, противоударное стекло от сколов и царапин на экран айфона Защитное стекло повторяет контуры экрана на 100% и закрывает его полностью от края до края, не оставляя зазоров. Благодаря наличию цветной рамки оно полностью копирует дизайн лицевой панели телефона и не портит его внешний вид. Комплектация Защитное стекло для iPhone 15/ Айфон 15 Спиртовая салфетка Салфетка из микрофибры Стикеры для удаления пыли Инструкция по наклеиванию Надежная упаковка', 'rating': 4.9, 'review_count': 229}
0
- Loss:
ContrastiveLoss
with these parameters:{ "distance_metric": "SiameseDistanceMetric.COSINE_DISTANCE", "margin": 0.5, "size_average": true }
Training Hyperparameters
Non-Default Hyperparameters
eval_strategy
: stepsnum_train_epochs
: 5warmup_ratio
: 0.1fp16
: Trueload_best_model_at_end
: Truebatch_sampler
: no_duplicates
All Hyperparameters
Click to expand
overwrite_output_dir
: Falsedo_predict
: Falseeval_strategy
: stepsprediction_loss_only
: Trueper_device_train_batch_size
: 8per_device_eval_batch_size
: 8per_gpu_train_batch_size
: Noneper_gpu_eval_batch_size
: Nonegradient_accumulation_steps
: 1eval_accumulation_steps
: Nonetorch_empty_cache_steps
: Nonelearning_rate
: 5e-05weight_decay
: 0.0adam_beta1
: 0.9adam_beta2
: 0.999adam_epsilon
: 1e-08max_grad_norm
: 1.0num_train_epochs
: 5max_steps
: -1lr_scheduler_type
: linearlr_scheduler_kwargs
: {}warmup_ratio
: 0.1warmup_steps
: 0log_level
: passivelog_level_replica
: warninglog_on_each_node
: Truelogging_nan_inf_filter
: Truesave_safetensors
: Truesave_on_each_node
: Falsesave_only_model
: Falserestore_callback_states_from_checkpoint
: Falseno_cuda
: Falseuse_cpu
: Falseuse_mps_device
: Falseseed
: 42data_seed
: Nonejit_mode_eval
: Falseuse_ipex
: Falsebf16
: Falsefp16
: Truefp16_opt_level
: O1half_precision_backend
: autobf16_full_eval
: Falsefp16_full_eval
: Falsetf32
: Nonelocal_rank
: 0ddp_backend
: Nonetpu_num_cores
: Nonetpu_metrics_debug
: Falsedebug
: []dataloader_drop_last
: Falsedataloader_num_workers
: 0dataloader_prefetch_factor
: Nonepast_index
: -1disable_tqdm
: Falseremove_unused_columns
: Truelabel_names
: Noneload_best_model_at_end
: Trueignore_data_skip
: Falsefsdp
: []fsdp_min_num_params
: 0fsdp_config
: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}fsdp_transformer_layer_cls_to_wrap
: Noneaccelerator_config
: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}deepspeed
: Nonelabel_smoothing_factor
: 0.0optim
: adamw_torchoptim_args
: Noneadafactor
: Falsegroup_by_length
: Falselength_column_name
: lengthddp_find_unused_parameters
: Noneddp_bucket_cap_mb
: Noneddp_broadcast_buffers
: Falsedataloader_pin_memory
: Truedataloader_persistent_workers
: Falseskip_memory_metrics
: Trueuse_legacy_prediction_loop
: Falsepush_to_hub
: Falseresume_from_checkpoint
: Nonehub_model_id
: Nonehub_strategy
: every_savehub_private_repo
: Nonehub_always_push
: Falsegradient_checkpointing
: Falsegradient_checkpointing_kwargs
: Noneinclude_inputs_for_metrics
: Falseinclude_for_metrics
: []eval_do_concat_batches
: Truefp16_backend
: autopush_to_hub_model_id
: Nonepush_to_hub_organization
: Nonemp_parameters
:auto_find_batch_size
: Falsefull_determinism
: Falsetorchdynamo
: Noneray_scope
: lastddp_timeout
: 1800torch_compile
: Falsetorch_compile_backend
: Nonetorch_compile_mode
: Nonedispatch_batches
: Nonesplit_batches
: Noneinclude_tokens_per_second
: Falseinclude_num_input_tokens_seen
: Falseneftune_noise_alpha
: Noneoptim_target_modules
: Nonebatch_eval_metrics
: Falseeval_on_start
: Falseuse_liger_kernel
: Falseeval_use_gather_object
: Falseaverage_tokens_across_devices
: Falseprompts
: Nonebatch_sampler
: no_duplicatesmulti_dataset_batch_sampler
: proportional
Training Logs
Click to expand
Epoch | Step | Training Loss | Validation Loss | item-classification_cosine_ap |
---|---|---|---|---|
0 | 0 | - | 0.0241 | 0.3377 |
0.0164 | 100 | 0.0182 | - | - |
0.0327 | 200 | 0.0137 | - | - |
0.0409 | 250 | - | 0.0128 | 0.4945 |
0.0491 | 300 | 0.0135 | - | - |
0.0655 | 400 | 0.0132 | - | - |
0.0818 | 500 | 0.0098 | 0.0092 | 0.7161 |
0.0982 | 600 | 0.0084 | - | - |
0.1146 | 700 | 0.0084 | - | - |
0.1228 | 750 | - | 0.0059 | 0.7971 |
0.1310 | 800 | 0.0074 | - | - |
0.1473 | 900 | 0.0072 | - | - |
0.1637 | 1000 | 0.0059 | 0.0050 | 0.8319 |
0.1801 | 1100 | 0.0054 | - | - |
0.1964 | 1200 | 0.0052 | - | - |
0.2046 | 1250 | - | 0.0046 | 0.8753 |
0.2128 | 1300 | 0.0048 | - | - |
0.2292 | 1400 | 0.0046 | - | - |
0.2455 | 1500 | 0.0049 | 0.0043 | 0.9045 |
0.2619 | 1600 | 0.0049 | - | - |
0.2783 | 1700 | 0.0046 | - | - |
0.2865 | 1750 | - | 0.0039 | 0.9027 |
0.2946 | 1800 | 0.0046 | - | - |
0.3110 | 1900 | 0.0045 | - | - |
0.3274 | 2000 | 0.0046 | 0.0035 | 0.9127 |
0.3438 | 2100 | 0.0043 | - | - |
0.3601 | 2200 | 0.0049 | - | - |
0.3683 | 2250 | - | 0.0033 | 0.9300 |
0.3765 | 2300 | 0.0042 | - | - |
0.3929 | 2400 | 0.0032 | - | - |
0.4092 | 2500 | 0.0038 | 0.0031 | 0.9393 |
0.4256 | 2600 | 0.0034 | - | - |
0.4420 | 2700 | 0.0042 | - | - |
0.4502 | 2750 | - | 0.0030 | 0.9418 |
0.4583 | 2800 | 0.004 | - | - |
0.4747 | 2900 | 0.0042 | - | - |
0.4911 | 3000 | 0.004 | 0.0031 | 0.9551 |
0.5074 | 3100 | 0.0038 | - | - |
0.5238 | 3200 | 0.0041 | - | - |
0.5320 | 3250 | - | 0.0032 | 0.9451 |
0.5402 | 3300 | 0.0041 | - | - |
0.5566 | 3400 | 0.0037 | - | - |
0.5729 | 3500 | 0.0032 | 0.0028 | 0.9585 |
0.5893 | 3600 | 0.0032 | - | - |
0.6057 | 3700 | 0.003 | - | - |
0.6138 | 3750 | - | 0.0029 | 0.9531 |
0.6220 | 3800 | 0.0031 | - | - |
0.6384 | 3900 | 0.0027 | - | - |
0.6548 | 4000 | 0.0024 | 0.0027 | 0.9559 |
0.6711 | 4100 | 0.0031 | - | - |
0.6875 | 4200 | 0.0025 | - | - |
0.6957 | 4250 | - | 0.0027 | 0.9637 |
0.7039 | 4300 | 0.0032 | - | - |
0.7202 | 4400 | 0.0034 | - | - |
0.7366 | 4500 | 0.0026 | 0.0024 | 0.9679 |
0.7530 | 4600 | 0.0025 | - | - |
0.7694 | 4700 | 0.0034 | - | - |
0.7775 | 4750 | - | 0.0024 | 0.9699 |
0.7857 | 4800 | 0.0024 | - | - |
0.8021 | 4900 | 0.0034 | - | - |
0.8185 | 5000 | 0.0028 | 0.0025 | 0.9624 |
0.8348 | 5100 | 0.0036 | - | - |
0.8512 | 5200 | 0.0025 | - | - |
0.8594 | 5250 | - | 0.0024 | 0.9666 |
0.8676 | 5300 | 0.0034 | - | - |
0.8839 | 5400 | 0.0026 | - | - |
0.9003 | 5500 | 0.0032 | 0.0024 | 0.9673 |
0.9167 | 5600 | 0.0032 | - | - |
0.9330 | 5700 | 0.0043 | - | - |
0.9412 | 5750 | - | 0.0026 | 0.9662 |
0.9494 | 5800 | 0.0027 | - | - |
0.9658 | 5900 | 0.0024 | - | - |
0.9822 | 6000 | 0.0037 | 0.0025 | 0.9691 |
0.9985 | 6100 | 0.0028 | - | - |
1.0149 | 6200 | 0.0031 | - | - |
1.0231 | 6250 | - | 0.0023 | 0.9671 |
1.0313 | 6300 | 0.0029 | - | - |
1.0476 | 6400 | 0.003 | - | - |
1.0640 | 6500 | 0.0027 | 0.0021 | 0.9689 |
1.0804 | 6600 | 0.0033 | - | - |
1.0967 | 6700 | 0.0027 | - | - |
1.1049 | 6750 | - | 0.0021 | 0.9735 |
1.1131 | 6800 | 0.0029 | - | - |
1.1295 | 6900 | 0.0023 | - | - |
1.1459 | 7000 | 0.0026 | 0.0020 | 0.9733 |
1.1622 | 7100 | 0.0024 | - | - |
1.1786 | 7200 | 0.0029 | - | - |
1.1868 | 7250 | - | 0.0021 | 0.9711 |
1.1950 | 7300 | 0.0023 | - | - |
1.2113 | 7400 | 0.0024 | - | - |
1.2277 | 7500 | 0.0031 | 0.0021 | 0.9753 |
1.2441 | 7600 | 0.0026 | - | - |
1.2604 | 7700 | 0.0019 | - | - |
1.2686 | 7750 | - | 0.0020 | 0.9713 |
1.2768 | 7800 | 0.0029 | - | - |
1.2932 | 7900 | 0.0022 | - | - |
1.3095 | 8000 | 0.0032 | 0.0020 | 0.9753 |
1.3259 | 8100 | 0.0021 | - | - |
1.3423 | 8200 | 0.002 | - | - |
1.3505 | 8250 | - | 0.0020 | 0.9744 |
1.3587 | 8300 | 0.003 | - | - |
1.3750 | 8400 | 0.0027 | - | - |
1.3914 | 8500 | 0.0019 | 0.0020 | 0.9752 |
1.4078 | 8600 | 0.0022 | - | - |
1.4241 | 8700 | 0.002 | - | - |
1.4323 | 8750 | - | 0.0020 | 0.9742 |
1.4405 | 8800 | 0.0021 | - | - |
1.4569 | 8900 | 0.0023 | - | - |
1.4732 | 9000 | 0.0026 | 0.0019 | 0.9749 |
1.4896 | 9100 | 0.0018 | - | - |
1.5060 | 9200 | 0.0023 | - | - |
1.5142 | 9250 | - | 0.0019 | 0.9753 |
1.5223 | 9300 | 0.0026 | - | - |
1.5387 | 9400 | 0.0022 | - | - |
1.5551 | 9500 | 0.0027 | 0.0020 | 0.9772 |
1.5715 | 9600 | 0.002 | - | - |
1.5878 | 9700 | 0.0019 | - | - |
1.5960 | 9750 | - | 0.0020 | 0.9776 |
1.6042 | 9800 | 0.0018 | - | - |
1.6206 | 9900 | 0.0019 | - | - |
1.6369 | 10000 | 0.0016 | 0.0020 | 0.9775 |
1.6533 | 10100 | 0.0017 | - | - |
1.6697 | 10200 | 0.0017 | - | - |
1.6779 | 10250 | - | 0.0019 | 0.9766 |
1.6860 | 10300 | 0.0014 | - | - |
1.7024 | 10400 | 0.0019 | - | - |
1.7188 | 10500 | 0.0023 | 0.0020 | 0.9769 |
1.7351 | 10600 | 0.0023 | - | - |
1.7515 | 10700 | 0.0017 | - | - |
1.7597 | 10750 | - | 0.0019 | 0.9760 |
1.7679 | 10800 | 0.0022 | - | - |
1.7843 | 10900 | 0.0017 | - | - |
1.8006 | 11000 | 0.0023 | 0.0019 | 0.9820 |
1.8170 | 11100 | 0.0018 | - | - |
1.8334 | 11200 | 0.0024 | - | - |
1.8415 | 11250 | - | 0.0020 | 0.9797 |
1.8497 | 11300 | 0.0016 | - | - |
1.8661 | 11400 | 0.0023 | - | - |
1.8825 | 11500 | 0.002 | 0.0020 | 0.9799 |
1.8988 | 11600 | 0.0022 | - | - |
1.9152 | 11700 | 0.0018 | - | - |
1.9234 | 11750 | - | 0.0021 | 0.9797 |
1.9316 | 11800 | 0.0028 | - | - |
1.9479 | 11900 | 0.0022 | - | - |
1.9643 | 12000 | 0.0015 | 0.0021 | 0.9799 |
1.9807 | 12100 | 0.0026 | - | - |
1.9971 | 12200 | 0.0019 | - | - |
2.0052 | 12250 | - | 0.0020 | 0.9807 |
2.0134 | 12300 | 0.0022 | - | - |
2.0298 | 12400 | 0.0022 | - | - |
2.0462 | 12500 | 0.0023 | 0.0019 | 0.9773 |
2.0625 | 12600 | 0.0022 | - | - |
2.0789 | 12700 | 0.0024 | - | - |
2.0871 | 12750 | - | 0.0019 | 0.9802 |
2.0953 | 12800 | 0.0018 | - | - |
2.1116 | 12900 | 0.0019 | - | - |
2.1280 | 13000 | 0.0019 | 0.0018 | 0.9815 |
2.1444 | 13100 | 0.0019 | - | - |
2.1607 | 13200 | 0.0019 | - | - |
2.1689 | 13250 | - | 0.0018 | 0.9818 |
2.1771 | 13300 | 0.0023 | - | - |
2.1935 | 13400 | 0.0016 | - | - |
2.2099 | 13500 | 0.0014 | 0.0019 | 0.9811 |
2.2262 | 13600 | 0.0022 | - | - |
2.2426 | 13700 | 0.002 | - | - |
2.2508 | 13750 | - | 0.0018 | 0.9817 |
2.2590 | 13800 | 0.0015 | - | - |
2.2753 | 13900 | 0.0023 | - | - |
2.2917 | 14000 | 0.0017 | 0.0019 | 0.9795 |
2.3081 | 14100 | 0.0025 | - | - |
2.3244 | 14200 | 0.0017 | - | - |
2.3326 | 14250 | - | 0.0018 | 0.9818 |
2.3408 | 14300 | 0.0016 | - | - |
2.3572 | 14400 | 0.0019 | - | - |
2.3735 | 14500 | 0.0019 | 0.0018 | 0.9825 |
2.3899 | 14600 | 0.0018 | - | - |
2.4063 | 14700 | 0.0015 | - | - |
2.4145 | 14750 | - | 0.0018 | 0.9829 |
2.4227 | 14800 | 0.0017 | - | - |
2.4390 | 14900 | 0.0019 | - | - |
2.4554 | 15000 | 0.0019 | 0.0018 | 0.9795 |
2.4718 | 15100 | 0.0018 | - | - |
2.4881 | 15200 | 0.0012 | - | - |
2.4963 | 15250 | - | 0.0018 | 0.9795 |
2.5045 | 15300 | 0.0017 | - | - |
2.5209 | 15400 | 0.0019 | - | - |
2.5372 | 15500 | 0.0018 | 0.0019 | 0.9801 |
2.5536 | 15600 | 0.0018 | - | - |
2.5700 | 15700 | 0.0018 | - | - |
2.5782 | 15750 | - | 0.0018 | 0.9805 |
2.5863 | 15800 | 0.0014 | - | - |
2.6027 | 15900 | 0.0013 | - | - |
2.6191 | 16000 | 0.0012 | 0.0017 | 0.9817 |
2.6355 | 16100 | 0.0013 | - | - |
2.6518 | 16200 | 0.0011 | - | - |
2.6600 | 16250 | - | 0.0018 | 0.9812 |
2.6682 | 16300 | 0.0012 | - | - |
2.6846 | 16400 | 0.0009 | - | - |
2.7009 | 16500 | 0.0015 | 0.0018 | 0.9809 |
2.7173 | 16600 | 0.0015 | - | - |
2.7337 | 16700 | 0.0019 | - | - |
2.7419 | 16750 | - | 0.0018 | 0.9811 |
2.7500 | 16800 | 0.0014 | - | - |
2.7664 | 16900 | 0.0017 | - | - |
2.7828 | 17000 | 0.001 | 0.0018 | 0.9817 |
2.7991 | 17100 | 0.0016 | - | - |
2.8155 | 17200 | 0.0014 | - | - |
2.8237 | 17250 | - | 0.0019 | 0.9829 |
2.8319 | 17300 | 0.0017 | - | - |
2.8483 | 17400 | 0.0012 | - | - |
2.8646 | 17500 | 0.0014 | 0.0018 | 0.9820 |
2.8810 | 17600 | 0.0014 | - | - |
2.8974 | 17700 | 0.0017 | - | - |
2.9055 | 17750 | - | 0.0018 | 0.9822 |
2.9137 | 17800 | 0.0016 | - | - |
2.9301 | 17900 | 0.0017 | - | - |
2.9465 | 18000 | 0.0018 | 0.0018 | 0.9818 |
2.9628 | 18100 | 0.0011 | - | - |
2.9792 | 18200 | 0.0019 | - | - |
2.9874 | 18250 | - | 0.0018 | 0.9817 |
2.9956 | 18300 | 0.0014 | - | - |
3.0119 | 18400 | 0.0017 | - | - |
3.0283 | 18500 | 0.0016 | 0.0017 | 0.9827 |
3.0447 | 18600 | 0.0015 | - | - |
3.0611 | 18700 | 0.0014 | - | - |
3.0692 | 18750 | - | 0.0017 | 0.9833 |
3.0774 | 18800 | 0.0021 | - | - |
3.0938 | 18900 | 0.0013 | - | - |
3.1102 | 19000 | 0.0012 | 0.0018 | 0.9844 |
3.1265 | 19100 | 0.0017 | - | - |
3.1429 | 19200 | 0.0015 | - | - |
3.1511 | 19250 | - | 0.0017 | 0.9840 |
3.1593 | 19300 | 0.0015 | - | - |
3.1756 | 19400 | 0.0017 | - | - |
3.1920 | 19500 | 0.0011 | 0.0017 | 0.9831 |
3.2084 | 19600 | 0.001 | - | - |
3.2248 | 19700 | 0.0014 | - | - |
3.2329 | 19750 | - | 0.0017 | 0.9836 |
3.2411 | 19800 | 0.0016 | - | - |
3.2575 | 19900 | 0.0013 | - | - |
3.2739 | 20000 | 0.0017 | 0.0017 | 0.9824 |
3.2902 | 20100 | 0.0013 | - | - |
3.3066 | 20200 | 0.002 | - | - |
3.3148 | 20250 | - | 0.0017 | 0.9813 |
3.3230 | 20300 | 0.0015 | - | - |
3.3393 | 20400 | 0.0011 | - | - |
3.3557 | 20500 | 0.0016 | 0.0017 | 0.9812 |
3.3721 | 20600 | 0.0016 | - | - |
3.3884 | 20700 | 0.0015 | - | - |
3.3966 | 20750 | - | 0.0017 | 0.9825 |
3.4048 | 20800 | 0.0012 | - | - |
3.4212 | 20900 | 0.0012 | - | - |
3.4376 | 21000 | 0.001 | 0.0017 | 0.9812 |
3.4539 | 21100 | 0.0019 | - | - |
3.4703 | 21200 | 0.0014 | - | - |
3.4785 | 21250 | - | 0.0017 | 0.9816 |
3.4867 | 21300 | 0.0009 | - | - |
3.5030 | 21400 | 0.0012 | - | - |
3.5194 | 21500 | 0.0015 | 0.0018 | 0.9823 |
3.5358 | 21600 | 0.0014 | - | - |
3.5521 | 21700 | 0.0015 | - | - |
3.5603 | 21750 | - | 0.0018 | 0.9814 |
3.5685 | 21800 | 0.0011 | - | - |
3.5849 | 21900 | 0.0012 | - | - |
3.6012 | 22000 | 0.001 | 0.0017 | 0.9822 |
3.6176 | 22100 | 0.0012 | - | - |
3.6340 | 22200 | 0.0009 | - | - |
3.6422 | 22250 | - | 0.0017 | 0.9823 |
3.6504 | 22300 | 0.0011 | - | - |
3.6667 | 22400 | 0.001 | - | - |
3.6831 | 22500 | 0.0008 | 0.0016 | 0.9825 |
3.6995 | 22600 | 0.0011 | - | - |
3.7158 | 22700 | 0.0014 | - | - |
3.7240 | 22750 | - | 0.0017 | 0.9826 |
3.7322 | 22800 | 0.0015 | - | - |
3.7486 | 22900 | 0.001 | - | - |
3.7649 | 23000 | 0.001 | 0.0017 | 0.9822 |
3.7813 | 23100 | 0.001 | - | - |
3.7977 | 23200 | 0.0014 | - | - |
3.8059 | 23250 | - | 0.0017 | 0.9836 |
3.8140 | 23300 | 0.0009 | - | - |
3.8304 | 23400 | 0.0013 | - | - |
3.8468 | 23500 | 0.001 | 0.0017 | 0.9845 |
3.8632 | 23600 | 0.001 | - | - |
3.8795 | 23700 | 0.001 | - | - |
3.8877 | 23750 | - | 0.0017 | 0.9848 |
3.8959 | 23800 | 0.0014 | - | - |
3.9123 | 23900 | 0.0017 | - | - |
3.9286 | 24000 | 0.0011 | 0.0017 | 0.9845 |
3.9450 | 24100 | 0.0014 | - | - |
3.9614 | 24200 | 0.0009 | - | - |
3.9696 | 24250 | - | 0.0019 | 0.9851 |
3.9777 | 24300 | 0.0015 | - | - |
3.9941 | 24400 | 0.0014 | - | - |
4.0105 | 24500 | 0.0013 | 0.0017 | 0.9862 |
4.0268 | 24600 | 0.0011 | - | - |
4.0432 | 24700 | 0.0014 | - | - |
4.0514 | 24750 | - | 0.0016 | 0.9848 |
4.0596 | 24800 | 0.0012 | - | - |
4.0760 | 24900 | 0.0014 | - | - |
4.0923 | 25000 | 0.0013 | 0.0017 | 0.9857 |
4.1087 | 25100 | 0.0008 | - | - |
4.1251 | 25200 | 0.0011 | - | - |
4.1332 | 25250 | - | 0.0017 | 0.9858 |
4.1414 | 25300 | 0.0013 | - | - |
4.1578 | 25400 | 0.0012 | - | - |
4.1742 | 25500 | 0.0012 | 0.0017 | 0.9858 |
4.1905 | 25600 | 0.0013 | - | - |
4.2069 | 25700 | 0.0008 | - | - |
4.2151 | 25750 | - | 0.0017 | 0.9855 |
4.2233 | 25800 | 0.0009 | - | - |
4.2396 | 25900 | 0.0012 | - | - |
4.2560 | 26000 | 0.0011 | 0.0016 | 0.9849 |
4.2724 | 26100 | 0.0015 | - | - |
4.2888 | 26200 | 0.0009 | - | - |
4.2969 | 26250 | - | 0.0017 | 0.9844 |
4.3051 | 26300 | 0.0013 | - | - |
4.3215 | 26400 | 0.0011 | - | - |
4.3379 | 26500 | 0.001 | 0.0017 | 0.9844 |
4.3542 | 26600 | 0.0014 | - | - |
4.3706 | 26700 | 0.0012 | - | - |
4.3788 | 26750 | - | 0.0016 | 0.9841 |
4.3870 | 26800 | 0.0013 | - | - |
4.4033 | 26900 | 0.0011 | - | - |
4.4197 | 27000 | 0.001 | 0.0016 | 0.9845 |
4.4361 | 27100 | 0.0008 | - | - |
4.4524 | 27200 | 0.0016 | - | - |
4.4606 | 27250 | - | 0.0016 | 0.9839 |
4.4688 | 27300 | 0.0011 | - | - |
4.4852 | 27400 | 0.0008 | - | - |
4.5016 | 27500 | 0.0009 | 0.0016 | 0.9847 |
4.5179 | 27600 | 0.0014 | - | - |
4.5343 | 27700 | 0.0011 | - | - |
4.5425 | 27750 | - | 0.0017 | 0.9849 |
4.5507 | 27800 | 0.0011 | - | - |
4.5670 | 27900 | 0.0008 | - | - |
4.5834 | 28000 | 0.001 | 0.0016 | 0.9846 |
4.5998 | 28100 | 0.0008 | - | - |
4.6161 | 28200 | 0.0008 | - | - |
4.6243 | 28250 | - | 0.0016 | 0.9839 |
4.6325 | 28300 | 0.0008 | - | - |
4.6489 | 28400 | 0.0007 | - | - |
4.6652 | 28500 | 0.0007 | 0.0016 | 0.9843 |
4.6816 | 28600 | 0.0008 | - | - |
4.6980 | 28700 | 0.0008 | - | - |
4.7062 | 28750 | - | 0.0016 | 0.9843 |
4.7144 | 28800 | 0.0011 | - | - |
4.7307 | 28900 | 0.0014 | - | - |
4.7471 | 29000 | 0.0008 | 0.0016 | 0.9841 |
4.7635 | 29100 | 0.0009 | - | - |
4.7798 | 29200 | 0.0006 | - | - |
4.7880 | 29250 | - | 0.0016 | 0.9840 |
4.7962 | 29300 | 0.001 | - | - |
4.8126 | 29400 | 0.0006 | - | - |
4.8289 | 29500 | 0.0013 | 0.0016 | 0.9843 |
4.8453 | 29600 | 0.0007 | - | - |
4.8617 | 29700 | 0.0008 | - | - |
4.8699 | 29750 | - | 0.0016 | 0.9844 |
4.8780 | 29800 | 0.001 | - | - |
4.8944 | 29900 | 0.0011 | - | - |
4.9108 | 30000 | 0.0013 | 0.0016 | 0.9846 |
4.9272 | 30100 | 0.001 | - | - |
4.9435 | 30200 | 0.0012 | - | - |
4.9517 | 30250 | - | 0.0017 | 0.9848 |
4.9599 | 30300 | 0.0007 | - | - |
4.9763 | 30400 | 0.001 | - | - |
4.9926 | 30500 | 0.0011 | 0.0017 | 0.9849 |
5.0 | 30545 | - | 0.0016 | 0.9839 |
- The bold row denotes the saved checkpoint.
Framework Versions
- Python: 3.10.13
- Sentence Transformers: 3.3.1
- Transformers: 4.47.1
- PyTorch: 2.2.1
- Accelerate: 1.2.1
- Datasets: 3.2.0
- Tokenizers: 0.21.0
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",
}
ContrastiveLoss
@inproceedings{hadsell2006dimensionality,
author={Hadsell, R. and Chopra, S. and LeCun, Y.},
booktitle={2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06)},
title={Dimensionality Reduction by Learning an Invariant Mapping},
year={2006},
volume={2},
number={},
pages={1735-1742},
doi={10.1109/CVPR.2006.100}
}
- Downloads last month
- 3
Model tree for denis-gordeev/reranker_dialog_items_biencoder_rubert-tiny-turbo-5
Evaluation results
- Cosine Accuracy on item classificationself-reported0.984
- Cosine Accuracy Threshold on item classificationself-reported0.725
- Cosine F1 on item classificationself-reported0.949
- Cosine F1 Threshold on item classificationself-reported0.725
- Cosine Precision on item classificationself-reported0.930
- Cosine Recall on item classificationself-reported0.970
- Cosine Ap on item classificationself-reported0.984