llazio's picture
Add new SentenceTransformer model.
d9b726d verified
metadata
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:1056
  - loss:BatchSemiHardTripletLoss
base_model: BAAI/bge-base-en
widget:
  - source_sentence: |

      Name : LearnTech Innovations
      Category: Educational Software, Professional Development Solutions
      Department: IT Operations
      Location: Tokyo, Japan
      Amount: 1523.45
      Card: Technology Faculty Enhancement Series
      Trip Name: unknown
    sentences:
      - |

        Name : OptiNet Solutions
        Category: Strategic Infrastructure Consulting, Advanced Data Solutions
        Department: Engineering
        Location: London, UK
        Amount: 1289.99
        Card: Next-Gen Network Design
        Trip Name: unknown
      - |

        Name : SkillsBox Academy
        Category: Online Education, Employee Training Platforms
        Department: Engineering
        Location: Chicago, IL
        Amount: 1298.5
        Card: Skill Development for New Technologies
        Trip Name: unknown
      - |

        Name : Yue Hua
        Category: HR & Employment Services
        Department: Engineering
        Location: Berlin, Germany
        Amount: 3567.45
        Card: Talent Acquisition Enhancement
        Trip Name: unknown
  - source_sentence: |

      Name : Interlink Global Solutions
      Category: Training Platforms, Professional Networking Services
      Department: HR
      Location: San Francisco, CA
      Amount: 1342.55
      Card: Executive Leadership Development
      Trip Name: unknown
    sentences:
      - |

        Name : Quantifire Insights
        Category: Predictive Analytics Solutions
        Department: Marketing
        Location: Zurich, Switzerland
        Amount: 1275.58
        Card: Customer Engagement Enhancement
        Trip Name: unknown
      - |

        Name : Insight Analytics
        Category: Data Services & Analytics, Employee Training & Development
        Department: Marketing
        Location: Berlin, Germany
        Amount: 873.29
        Card: Market Research Initiative
        Trip Name: unknown
      - |

        Name : GlobalSafe Solutions
        Category: Risk Management Consultancy, International Insurance Brokerage
        Department: Finance
        Location: Zurich, Switzerland
        Amount: 1175.65
        Card: Annual Risk Assessment
        Trip Name: unknown
  - source_sentence: |

      Name : SkyScale Services
      Category: Global IT Solutions, Platform Integration
      Department: IT Operations
      Location: Dublin, Ireland
      Amount: 1492.54
      Card: Annual Platform Enhancement
      Trip Name: unknown
    sentences:
      - |

        Name : InterGlobe Connect
        Category: Financial Services, Data Connectivity
        Department: Finance
        Location: London, UK
        Amount: 152.79
        Card: Overseas Financial Operations
        Trip Name: unknown
      - |

        Name : SkyElevate Group
        Category: Luxury Travel Services, Corporate Event Planning
        Department: Executive
        Location: Dubai, UAE
        Amount: 2113.47
        Card: Executive Strategy Retreat
        Trip Name: Board of Directors Retreat
      - |

        Name : GlobalRes Workforce Solutions
        Category: Remote Work Platforms, HR Technology Vendors
        Department: Engineering
        Location: Barcelona, Spain
        Amount: 1894.27
        Card: Hybrid Work Enablement
        Trip Name: unknown
  - source_sentence: |

      Name : Nimbus Networks Inc.
      Category: Cloud Services, Application Hosting
      Department: Research & Development
      Location: Austin, TX
      Amount: 1134.67
      Card: NextGen Application Deployment
      Trip Name: unknown
    sentences:
      - >

        Name : Kaleidoscope Interactive

        Category: Interactive Software Platforms, Educational Content
        Distribution

        Department: Engineering

        Location: London, UK

        Amount: 1852.37

        Card: Innovative Education Initiative

        Trip Name: unknown
      - |

        Name : RBS
        Category: Financial Services, Business Consultancy
        Department: Finance
        Location: Toronto, Canada
        Amount: 1134.28
        Card: Cross-Border Transaction Facilitation
        Trip Name: unknown
      - |

        Name : HexaGuard Systems
        Category: Enterprise Software Solutions, Network Infrastructure Services
        Department: IT Operations
        Location: Toronto, Canada
        Amount: 1254.78
        Card: Integrated Security Enhancement Plan
        Trip Name: unknown
  - source_sentence: |

      Name : SMRT
      Category: Public Transportation, Transit Services
      Department: Sales
      Location: Los Angeles, CA
      Amount: 85.0
      Card: Sales Team Travel Budget
      Trip Name: unknown
    sentences:
      - |

        Name : Wellness Haven
        Category: Employee Health Programs, Professional Development
        Department: HR
        Location: Munich, Germany
        Amount: 762.35
        Card: Corporate Wellness Initiatives
        Trip Name: unknown
      - |

        Name : Elastic Habitat Solutions
        Category: Cloud Enhancement Services, Data Analytics Tools
        Department: IT Operations
        Location: London, UK
        Amount: 1489.57
        Card: Scalable Data Initiative
        Trip Name: unknown
      - |

        Name : NexaCloud Technologies
        Category: Implement Services, Cloud Solutions
        Department: IT Operations
        Location: Berlin, Germany
        Amount: 1490.65
        Card: Cloud Optimization Initiative
        Trip Name: unknown
pipeline_tag: sentence-similarity
library_name: sentence-transformers
metrics:
  - cosine_accuracy
  - dot_accuracy
  - manhattan_accuracy
  - euclidean_accuracy
  - max_accuracy
model-index:
  - name: SentenceTransformer based on BAAI/bge-base-en
    results:
      - task:
          type: triplet
          name: Triplet
        dataset:
          name: bge base en train
          type: bge-base-en-train
        metrics:
          - type: cosine_accuracy
            value: 0.9753787878787878
            name: Cosine Accuracy
          - type: dot_accuracy
            value: 0.02462121212121212
            name: Dot Accuracy
          - type: manhattan_accuracy
            value: 0.9753787878787878
            name: Manhattan Accuracy
          - type: euclidean_accuracy
            value: 0.9753787878787878
            name: Euclidean Accuracy
          - type: max_accuracy
            value: 0.9753787878787878
            name: Max Accuracy
      - task:
          type: triplet
          name: Triplet
        dataset:
          name: bge base en eval
          type: bge-base-en-eval
        metrics:
          - type: cosine_accuracy
            value: 0.9813432835820896
            name: Cosine Accuracy
          - type: dot_accuracy
            value: 0.018656716417910446
            name: Dot Accuracy
          - type: manhattan_accuracy
            value: 0.9850746268656716
            name: Manhattan Accuracy
          - type: euclidean_accuracy
            value: 0.9813432835820896
            name: Euclidean Accuracy
          - type: max_accuracy
            value: 0.9850746268656716
            name: Max Accuracy

SentenceTransformer based on BAAI/bge-base-en

This is a sentence-transformers model finetuned from BAAI/bge-base-en. 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: BAAI/bge-base-en
  • Maximum Sequence Length: 512 tokens
  • Output Dimensionality: 768 tokens
  • Similarity Function: Cosine Similarity

Model Sources

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 512, 'do_lower_case': True}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 768, '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("llazio/finetuned-bge-base-en")
# Run inference
sentences = [
    '\nName : SMRT\nCategory: Public Transportation, Transit Services\nDepartment: Sales\nLocation: Los Angeles, CA\nAmount: 85.0\nCard: Sales Team Travel Budget\nTrip Name: unknown\n',
    '\nName : Elastic Habitat Solutions\nCategory: Cloud Enhancement Services, Data Analytics Tools\nDepartment: IT Operations\nLocation: London, UK\nAmount: 1489.57\nCard: Scalable Data Initiative\nTrip Name: unknown\n',
    '\nName : Wellness Haven\nCategory: Employee Health Programs, Professional Development\nDepartment: HR\nLocation: Munich, Germany\nAmount: 762.35\nCard: Corporate Wellness Initiatives\nTrip Name: unknown\n',
]
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

Triplet

Metric Value
cosine_accuracy 0.9754
dot_accuracy 0.0246
manhattan_accuracy 0.9754
euclidean_accuracy 0.9754
max_accuracy 0.9754

Triplet

Metric Value
cosine_accuracy 0.9813
dot_accuracy 0.0187
manhattan_accuracy 0.9851
euclidean_accuracy 0.9813
max_accuracy 0.9851

Training Details

Training Dataset

Unnamed Dataset

  • Size: 1,056 training samples
  • Columns: sentence and label
  • Approximate statistics based on the first 1000 samples:
    sentence label
    type string int
    details
    • min: 32 tokens
    • mean: 39.65 tokens
    • max: 49 tokens
    • 0: ~4.80%
    • 1: ~4.80%
    • 2: ~5.60%
    • 3: ~2.20%
    • 4: ~5.10%
    • 5: ~4.10%
    • 6: ~3.70%
    • 7: ~4.30%
    • 8: ~4.80%
    • 9: ~4.50%
    • 10: ~1.80%
    • 11: ~2.60%
    • 12: ~4.50%
    • 13: ~5.00%
    • 14: ~4.20%
    • 15: ~2.70%
    • 16: ~3.50%
    • 17: ~4.10%
    • 18: ~2.70%
    • 19: ~3.40%
    • 20: ~3.80%
    • 21: ~3.70%
    • 22: ~4.20%
    • 23: ~1.80%
    • 24: ~2.60%
    • 25: ~2.50%
    • 26: ~3.00%
  • Samples:
    sentence label

    Name : Global Talent Network
    Category: HR Consultancy Services, Corporate Event Organizers
    Department: HR
    Location: Los Angeles, CA
    Amount: 1375.65
    Card: Leadership Summit Coordination
    Trip Name: unknown
    0

    Name : Baku
    Category: Ride Sharing
    Department: Sales
    Location: Baku, Azerbaijan
    Amount: 1247.88
    Card: Client Engagement Activities
    Trip Name: unknown
    1

    Name : Harris Consulting Group
    Category: Business Consulting, Legal Advisory
    Department: Finance
    Location: Toronto, Canada
    Amount: 1325.45
    Card: Strategic Development Fund
    Trip Name: unknown
    2
  • Loss: BatchSemiHardTripletLoss

Evaluation Dataset

Unnamed Dataset

  • Size: 264 evaluation samples
  • Columns: sentence and label
  • Approximate statistics based on the first 264 samples:
    sentence label
    type string int
    details
    • min: 34 tokens
    • mean: 39.59 tokens
    • max: 49 tokens
    • 0: ~5.30%
    • 1: ~4.92%
    • 2: ~6.44%
    • 3: ~2.27%
    • 4: ~3.03%
    • 5: ~4.55%
    • 6: ~3.79%
    • 7: ~4.17%
    • 8: ~3.79%
    • 9: ~3.41%
    • 10: ~1.14%
    • 11: ~1.89%
    • 12: ~5.30%
    • 13: ~4.92%
    • 14: ~3.41%
    • 15: ~2.27%
    • 16: ~2.65%
    • 17: ~2.65%
    • 18: ~3.79%
    • 19: ~3.79%
    • 20: ~6.06%
    • 21: ~6.44%
    • 22: ~3.41%
    • 23: ~1.89%
    • 24: ~3.79%
    • 25: ~2.65%
    • 26: ~2.27%
  • Samples:
    sentence label

    Name : Fernández & Co. Services
    Category: Property Management, Facility Services
    Department: Office Administration
    Location: Madrid, Spain
    Amount: 1245.67
    Card: Monthly Facility Operations
    Trip Name: unknown
    18

    Name : Habitat Solutions
    Category: Cloud Enhancement Services, Data Analytics Tools
    Department: IT Operations
    Location: London, UK
    Amount: 1489.57
    Card: Scalable Data Initiative
    Trip Name: unknown
    13

    Name : Gandalf
    Category: Financial Services, Consulting
    Department: Finance
    Location: Singapore
    Amount: 457.29
    Card: Financial Advisory Services
    Trip Name: unknown
    21
  • Loss: BatchSemiHardTripletLoss

Training Hyperparameters

Non-Default Hyperparameters

  • eval_strategy: steps
  • per_device_train_batch_size: 16
  • per_device_eval_batch_size: 16
  • learning_rate: 2e-05
  • num_train_epochs: 5
  • warmup_ratio: 0.1
  • 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: 16
  • per_device_eval_batch_size: 16
  • 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: False
  • 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
  • use_liger_kernel: False
  • eval_use_gather_object: False
  • batch_sampler: no_duplicates
  • multi_dataset_batch_sampler: proportional

Training Logs

Epoch Step Training Loss loss bge-base-en-eval_max_accuracy bge-base-en-train_max_accuracy
0 0 - - - 0.8428
1.5152 100 4.9094 4.7659 - 0.9688
3.0303 200 4.7495 4.6519 - 0.9725
4.5455 300 4.6271 4.5820 - 0.9754
5.0 330 - - 0.9851 -

Framework Versions

  • Python: 3.9.6
  • Sentence Transformers: 3.1.1
  • Transformers: 4.45.2
  • PyTorch: 2.6.0
  • Accelerate: 1.3.0
  • Datasets: 3.2.0
  • Tokenizers: 0.20.3

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",
}

BatchSemiHardTripletLoss

@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}
}