gromoboy/qwen3_06b_items_matcher

SentenceTransformer based on Qwen/Qwen3-Embedding-0.6B

This is a sentence-transformers model finetuned from Qwen/Qwen3-Embedding-0.6B on the data1 dataset. 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: Qwen/Qwen3-Embedding-0.6B
• Maximum Sequence Length: 32768 tokens
• Output Dimensionality: 1024 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • data1

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': 32768, 'do_lower_case': False, 'architecture': 'Qwen3Model'})
  (1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': False, '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': True, '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("gromoboy/qwen3_06b_items_matcher")
# Run inference
queries = [
    "\u041e\u0434\u0435\u0436\u0434\u0430 \u0434\u043b\u044f \u043d\u043e\u0432\u043e\u0440\u043e\u0436\u0434\u0435\u043d\u043d\u044b\u0445 \u043c\u0430\u043b\u044c\u0447\u0438\u043a\u043e\u0432 \u0441\u043b\u0438\u043f \u0434\u043b\u044f \u043c\u0430\u043b\u044b\u0448\u0435\u0439 \u043a\u043e\u043c\u0431\u0438\u043d\u0435\u0437\u043e\u043d \u043d\u0430\u0440\u044f\u0434\u043d\u044b\u0439 \u043d\u0430\u0442\u0435\u043b\u044c\u043d\u044b\u0439 \u0434\u043b\u044f \u0444\u043e\u0442\u043e\u0441\u0435\u0441\u0441\u0438\u0438",
]
documents = [
    'Одежда для новорожденных мальчиков слипдля малышей комбинезон нарядный нательный для фотосесии',
    'Шапка детская для мальчика и снуд',
    'ТелескопРефрактор/Детский игровойнабор',
]
query_embeddings = model.encode_query(queries)
document_embeddings = model.encode_document(documents)
print(query_embeddings.shape, document_embeddings.shape)
# [1, 1024] [3, 1024]

# Get the similarity scores for the embeddings
similarities = model.similarity(query_embeddings, document_embeddings)
print(similarities)
# tensor([[0.9531, 0.2704, 0.1847]])

Evaluation

Metrics

Triplet

• Dataset: dev
• Evaluated with TripletEvaluator with these parameters:

  {
      "margin": {
          "cosine": 0.3,
          "dot": 0.3,
          "manhattan": 0.3,
          "euclidean": 0.3
      }
  }
  

Metric	Value
cosine_accuracy	0.9423

Training Details

Training Dataset

data1

• Dataset: data1
• Size: 8,914 training samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 6 tokens mean: 14.74 tokens max: 46 tokens	min: 36 tokens mean: 51.42 tokens max: 86 tokens	min: 6 tokens mean: 14.61 tokens max: 46 tokens

• Samples:

  anchor	positive	negative
  Cоуc рыбный Cook&Lobster	Соус рыбный, Таиланд 750мл*12 ,стекло (Штук/ящ: [12], Вес в кг: [1.448]	Соус устричный Genso
  Cоуc рыбный Cook&Lobster	Соус рыбный, Таиланд, 700мл*12 (Штук/ящ: [12], Вес в кг: [1.250]	Соус устричный Genso
  Kimchi Чипсы нори Tidori	Чипсы нори TIDORI, Корея, Kimchi, 15г (5г х 3) * 24 (Штук/ящ: [24], Вес в кг: [0.038]	Original Чипсы нори Tidori

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 25,
      "similarity_fct": "cos_sim"
  }
  

Evaluation Dataset

data1

• Dataset: data1
• Size: 2,288 evaluation samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 6 tokens mean: 19.04 tokens max: 83 tokens	min: 6 tokens mean: 34.31 tokens max: 88 tokens	min: 6 tokens mean: 18.38 tokens max: 100 tokens

• Samples:

  anchor	positive	negative
  BBQ Чипсы нори Tidori	Чипсы нори TIDORI, Корея, BBQ, 15г (5г х 3) * 24 (Штук/ящ: [24], Вес в кг: [0.038]	Kimchi Чипсы нори Tidori
  Original Чипсы нори Tidori	Чипсы нори TIDORI, Корея, Original, 15г (5г х 3) * 24 (Штук/ящ: [24], Вес в кг: [0.038]	Kimchi Чипсы нори Tidori
  Авокадо пюре десертное с кокосом, голубикой и сиропом агавы, быстрозамороженное, блок (57 г*4)	Авокадо пюре десерт. с КОКОСОМ, ГОЛУБИКОЙ и сиропом агавы, быстрозамороженный 227гр12 блок (57гр4) (Штук/ящ: [12], Вес в кг: [0.235]	Авокадо пюре с киви, мятой и сиропом агавы, быстрозамороженное, блок (57 г*4)

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 25,
      "similarity_fct": "cos_sim"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 32
• learning_rate: 2e-05
• num_train_epochs: 5
• warmup_ratio: 0.1
• fp16: True
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 32
• 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: 5
• 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}
• 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
• hub_revision: None
• 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
• 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
• liger_kernel_config: None
• eval_use_gather_object: False
• average_tokens_across_devices: False
• prompts: None
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional
• router_mapping: {}
• learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss	Validation Loss	dev_cosine_accuracy
-1	-1	-	-	0.5848
0.3584	100	-	0.0570	0.9030
0.7168	200	0.0638	0.0504	0.9008
1.0753	300	-	0.0431	0.9331
1.4337	400	0.0067	0.0385	0.9292
1.7921	500	-	0.0715	0.9191
2.1505	600	0.0045	0.0664	0.9309
2.5090	700	-	0.0620	0.9414
2.8674	800	0.0029	0.0532	0.9467
3.2258	900	-	0.0586	0.9432
3.5842	1000	0.0041	0.0431	0.9432
3.9427	1100	-	0.0464	0.9432
4.3011	1200	0.0022	0.0611	0.9406
4.6595	1300	-	0.0646	0.9423

Framework Versions

• Python: 3.11.10
• Sentence Transformers: 5.0.0
• Transformers: 4.54.0
• PyTorch: 2.5.1+cu124
• Accelerate: 1.9.0
• Datasets: 4.0.0
• Tokenizers: 0.21.2

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",
}

MultipleNegativesRankingLoss

@misc{henderson2017efficient,
    title={Efficient Natural Language Response Suggestion for Smart Reply},
    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
    year={2017},
    eprint={1705.00652},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}