SentenceTransformer based on deepvk/USER-bge-m3
This is a sentence-transformers model finetuned from deepvk/USER-bge-m3. It maps sentences & paragraphs to a 1024-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: deepvk/USER-bge-m3
- Maximum Sequence Length: 512 tokens
- Output Dimensionality: 1024 tokens
- 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: XLMRobertaModel
(1): Pooling({'word_embedding_dimension': 1024, '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("Data-Lab/USER-bge-m3-embedder-td")
# Run inference
sentences = [
'детская каша',
'Каша овсяная детская "Мишка" Сладкая овсяная каша с голубикой и бананами. Можно приготовить на кокосовом молоке',
'Десерт "Тирамису", 300 г Изысканный итальянский десерт в нестандартном исполнении. В нашем Тирамису много (очень много!) сливочного крема и Маскарпоне, поэтому лакомство невероятно нежное!',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]
# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]
Evaluation
Metrics
Triplet
- Dataset:
dev
- Evaluated with
TripletEvaluator
Metric | Value |
---|---|
cosine_accuracy | 0.9188 |
dot_accuracy | 0.0803 |
manhattan_accuracy | 0.917 |
euclidean_accuracy | 0.9188 |
max_accuracy | 0.9188 |
Training Details
Training Dataset
Unnamed Dataset
- Size: 10,189 training samples
- Columns:
sentence_0
,sentence_1
, andsentence_2
- Approximate statistics based on the first 1000 samples:
sentence_0 sentence_1 sentence_2 type string string string details - min: 3 tokens
- mean: 7.85 tokens
- max: 30 tokens
- min: 6 tokens
- mean: 61.74 tokens
- max: 377 tokens
- min: 5 tokens
- mean: 64.71 tokens
- max: 393 tokens
- Samples:
sentence_0 sentence_1 sentence_2 хурма
Чипсы из хурмы, 25 г Натуральные чипсы из хурмы, без сахара. Мягкие, медово-фруктовые
Салат мимоза, 300 г Классический салат мимоза с горбушей, отварными овощами и куриными желтками.
жареное мясо
КК_котлета куриная жареная, вес
Баклажаны "Пармиджано" Мама миа, это же настоящая итальянская пармиджана! Нежные ломтики баклажанов, много томатов и ещё больше тягучего сыра. Очень насыщенно, сочно и аппетитно пряно. Баклажаны для этого рецепта не обжариваются, а запекаются в духовке, что делает блюдо более полезным и изысканным.
бедро цыпленка бройлера
Бедро цыплят-бройлеров Халяль 1 кг Сочное бедро цыпленка, подходит для маринования, тушения и запекания
Мясо бедра (Филе бедра) индейки в маринаде "Чесночный" 1 кг Диетическое, нежирное филе бедра индейки с деликатным вкусом и ароматом. В меру подсолено и приправлено острым чесночком и травами.
- Loss:
TripletLoss
with these parameters:{ "distance_metric": "TripletDistanceMetric.COSINE", "triplet_margin": 0.5 }
Training Hyperparameters
Non-Default Hyperparameters
eval_strategy
: stepsper_device_train_batch_size
: 4per_device_eval_batch_size
: 4fp16
: Truemulti_dataset_batch_sampler
: round_robin
All Hyperparameters
Click to expand
overwrite_output_dir
: Falsedo_predict
: Falseeval_strategy
: stepsprediction_loss_only
: Trueper_device_train_batch_size
: 4per_device_eval_batch_size
: 4per_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
: 1num_train_epochs
: 3max_steps
: -1lr_scheduler_type
: linearlr_scheduler_kwargs
: {}warmup_ratio
: 0.0warmup_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
: Truedataloader_num_workers
: 0dataloader_prefetch_factor
: Nonepast_index
: -1disable_tqdm
: Falseremove_unused_columns
: Truelabel_names
: Noneload_best_model_at_end
: Falseignore_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
: Falsehub_always_push
: Falsegradient_checkpointing
: Falsegradient_checkpointing_kwargs
: Noneinclude_inputs_for_metrics
: Falseeval_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
: Falseeval_use_gather_object
: Falsebatch_sampler
: batch_samplermulti_dataset_batch_sampler
: round_robin
Training Logs
Epoch | Step | Training Loss | dev_max_accuracy |
---|---|---|---|
0.3928 | 500 | 0.2477 | - |
0.7855 | 1000 | 0.182 | 0.9064 |
1.0 | 1273 | - | 0.9073 |
1.1783 | 1500 | 0.157 | - |
1.5711 | 2000 | 0.1234 | 0.9029 |
1.9639 | 2500 | 0.0993 | - |
2.0 | 2546 | - | 0.9179 |
2.3566 | 3000 | 0.0864 | 0.9170 |
2.7494 | 3500 | 0.0691 | - |
3.0 | 3819 | - | 0.9188 |
Framework Versions
- Python: 3.10.12
- Sentence Transformers: 3.2.0
- Transformers: 4.44.0
- PyTorch: 2.3.1+cu121
- Accelerate: 0.31.0
- 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",
}
TripletLoss
@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}
}
- Downloads last month
- 54
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Model tree for Data-Lab/USER-bge-m3-embedder-td
Base model
deepvk/USER-bge-m3Evaluation results
- Cosine Accuracy on devself-reported0.919
- Dot Accuracy on devself-reported0.080
- Manhattan Accuracy on devself-reported0.917
- Euclidean Accuracy on devself-reported0.919
- Max Accuracy on devself-reported0.919