rdxtremity/harrir-bge-m3

SentenceTransformer based on BAAI/bge-m3

This is a sentence-transformers model finetuned from BAAI/bge-m3. It maps sentences & paragraphs to a 1024-dimensional dense vector space and can be used for retrieval.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: BAAI/bge-m3
• Maximum Sequence Length: 128 tokens
• Output Dimensionality: 1024 dimensions
• Similarity Function: Cosine Similarity
• Supported Modality: Text

Model Sources

• Documentation: Sentence Transformers Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Sentence Transformers on Hugging Face

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'transformer_task': 'feature-extraction', 'modality_config': {'text': {'method': 'forward', 'method_output_name': 'last_hidden_state'}}, 'module_output_name': 'token_embeddings', 'architecture': 'XLMRobertaModel'})
  (1): Pooling({'embedding_dimension': 1024, 'pooling_mode': 'cls', '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("sentence_transformers_model_id")
# Run inference
sentences = [
    'ردية ميدي نقوش زهور',
    '[PRETTY LAVISH] Floral Print Halter Maxi Dress | فستان مكسي هولتر بنقشة الزهور. Category: Dresses > Maxi Dresses.',
    '[Ginger] Boxy Cropped Shirt | قميص كروب بقصة مربعة. Category: Tops > Other Tops.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[0.9999, 0.4676, 0.4029],
#         [0.4676, 1.0000, 0.4774],
#         [0.4029, 0.4774, 1.0000]])

Evaluation

Metrics

Triplet

• Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.8765

Training Details

Training Dataset

Unnamed Dataset

• Size: 20,469 training samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 100 samples:

  	anchor	positive	negative
  type	string	string	string
  modality	text	text	text
  details	min: 4 tokens mean: 6.24 tokens max: 10 tokens	min: 29 tokens mean: 35.47 tokens max: 41 tokens	min: 24 tokens mean: 39.13 tokens max: 72 tokens

• Samples:

  anchor	positive	negative
  nike cortez	[LACOSTE] T-Clip Low-Top Court Sneakers | تي-كليب سنيكرز كورت منخفض. Category: Shoes > Sneakers.	[Nike] Nike Sportswear Knit Crop Tank | نايك توب تانك كروب محبوك سبورتس وير. Category: Tops > Tank Tops & Camis.
  nike cortez	[Nike] Nike Cortez TXT | نايك كورتيز قماش. Category: Shoes > Sneakers.	[Nike] Dri-Fit Metal Swoosh Cap Black | قبعة دراي-فيت كلوب. Category: Accessories > The Hat Store.
  nike cortez	[Nike] Nike Cortez TXT | نايك كورتيز قماش. Category: Shoes > Sneakers.	[TOMMY HILFIGER] Vulcanized Low Top Sneakers | سنيكرز فولكنايزد منخفض. Category: Shoes > Sneakers.

• Loss: TripletLoss with these parameters:

  {
      "distance_metric": "TripletDistanceMetric.COSINE",
      "triplet_margin": 0.2
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 16
• num_train_epochs: 1
• learning_rate: 5e-06
• warmup_steps: 0.1
• gradient_accumulation_steps: 4
• bf16: True
• tf32: True

All Hyperparameters

Click to expand

• per_device_train_batch_size: 16
• num_train_epochs: 1
• max_steps: -1
• learning_rate: 5e-06
• lr_scheduler_type: linear
• lr_scheduler_kwargs: None
• warmup_steps: 0.1
• optim: adamw_torch_fused
• optim_args: None
• weight_decay: 0.0
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• optim_target_modules: None
• gradient_accumulation_steps: 4
• average_tokens_across_devices: True
• max_grad_norm: 1.0
• label_smoothing_factor: 0.0
• bf16: True
• fp16: False
• bf16_full_eval: False
• fp16_full_eval: False
• tf32: True
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• torch_compile: False
• torch_compile_backend: None
• torch_compile_mode: None
• use_liger_kernel: False
• liger_kernel_config: None
• use_cache: False
• neftune_noise_alpha: None
• torch_empty_cache_steps: None
• auto_find_batch_size: False
• log_on_each_node: True
• logging_nan_inf_filter: True
• include_num_input_tokens_seen: no
• log_level: passive
• log_level_replica: warning
• disable_tqdm: False
• project: huggingface
• trackio_space_id: None
• trackio_bucket_id: None
• trackio_static_space_id: None
• per_device_eval_batch_size: 8
• prediction_loss_only: True
• eval_on_start: False
• eval_do_concat_batches: True
• eval_use_gather_object: False
• eval_accumulation_steps: None
• include_for_metrics: []
• batch_eval_metrics: False
• save_only_model: False
• save_on_each_node: False
• enable_jit_checkpoint: False
• push_to_hub: False
• hub_private_repo: None
• hub_model_id: None
• hub_strategy: every_save
• hub_always_push: False
• hub_revision: None
• load_best_model_at_end: False
• ignore_data_skip: False
• restore_callback_states_from_checkpoint: False
• full_determinism: False
• seed: 42
• data_seed: None
• use_cpu: False
• accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
• parallelism_config: None
• dataloader_drop_last: False
• dataloader_num_workers: 0
• dataloader_pin_memory: True
• dataloader_persistent_workers: False
• dataloader_prefetch_factor: None
• remove_unused_columns: True
• label_names: None
• train_sampling_strategy: random
• length_column_name: length
• ddp_find_unused_parameters: None
• ddp_bucket_cap_mb: None
• ddp_broadcast_buffers: False
• ddp_static_graph: None
• ddp_backend: None
• ddp_timeout: 1800
• fsdp: []
• fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
• deepspeed: None
• debug: []
• skip_memory_metrics: True
• do_predict: False
• resume_from_checkpoint: None
• warmup_ratio: None
• local_rank: -1
• prompts: None
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: proportional
• router_mapping: {}
• learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss	cosine_accuracy
-1	-1	-	0.8667
0.0312	10	0.1282	-
0.0625	20	0.1300	-
0.0938	30	0.1292	-
0.125	40	0.1279	-
0.1562	50	0.1285	-
0.1875	60	0.1326	-
0.2188	70	0.1279	-
0.25	80	0.1220	-
0.2812	90	0.1263	-
0.3125	100	0.1231	-
0.3438	110	0.1227	-
0.375	120	0.1200	-
0.4062	130	0.1257	-
0.4375	140	0.1194	-
0.4688	150	0.1237	-
0.5	160	0.1248	-
0.5312	170	0.1220	-
0.5625	180	0.1208	-
0.5938	190	0.1199	-
0.625	200	0.1208	-
0.6562	210	0.1197	-
0.6875	220	0.1156	-
0.7188	230	0.1186	-
0.75	240	0.1227	-
0.7812	250	0.1171	-
0.8125	260	0.1214	-
0.8438	270	0.1156	-
0.875	280	0.1203	-
0.9062	290	0.1210	-
0.9375	300	0.1164	-
0.9688	310	0.1164	-
1.0	320	0.1213	-
-1	-1	-	0.8765

Training Time

• Training: 2.5 minutes

Framework Versions

• Python: 3.11.12
• Sentence Transformers: 5.5.0
• Transformers: 5.8.1
• PyTorch: 2.12.0+cu130
• Accelerate: 1.13.0
• Datasets: 4.8.5
• Tokenizers: 0.22.2

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",
}

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}
}