bhlim/bge-base-financial-matryoshka

BGE base Financial Matryoshka

This is a sentence-transformers model finetuned from BAAI/bge-base-en-v1.5. 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-v1.5
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 tokens
• Similarity Function: Cosine Similarity
• Language: en
• License: apache-2.0

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': 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("bhlim/bge-base-financial-matryoshka")
# Run inference
sentences = [
    'Our revenue consists of service fees, net of incentives and refunds, charged to our customers. For stays, service fees, which are charged to customers as a percentage of the value of the booking, excluding taxes, vary based on factors specific to the booking, such as booking value, the duration of the booking, geography, and Host type.',
    'What are some factors that affect the percentage of service fees charged to customers?',
    "What is the PCAOB ID number for PricewaterhouseCoopers LLP concerning the firm's financial statements?",
]
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

Information Retrieval

• Dataset: dim_768
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6957
cosine_accuracy@3	0.8
cosine_accuracy@5	0.8486
cosine_accuracy@10	0.9
cosine_precision@1	0.6957
cosine_precision@3	0.2667
cosine_precision@5	0.1697
cosine_precision@10	0.09
cosine_recall@1	0.6957
cosine_recall@3	0.8
cosine_recall@5	0.8486
cosine_recall@10	0.9
cosine_ndcg@10	0.7935
cosine_mrr@10	0.76
cosine_map@100	0.764

Information Retrieval

• Dataset: dim_512
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.7057
cosine_accuracy@3	0.8014
cosine_accuracy@5	0.8529
cosine_accuracy@10	0.9029
cosine_precision@1	0.7057
cosine_precision@3	0.2671
cosine_precision@5	0.1706
cosine_precision@10	0.0903
cosine_recall@1	0.7057
cosine_recall@3	0.8014
cosine_recall@5	0.8529
cosine_recall@10	0.9029
cosine_ndcg@10	0.7984
cosine_mrr@10	0.7656
cosine_map@100	0.7693

Information Retrieval

• Dataset: dim_256
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6914
cosine_accuracy@3	0.79
cosine_accuracy@5	0.8471
cosine_accuracy@10	0.8914
cosine_precision@1	0.6914
cosine_precision@3	0.2633
cosine_precision@5	0.1694
cosine_precision@10	0.0891
cosine_recall@1	0.6914
cosine_recall@3	0.79
cosine_recall@5	0.8471
cosine_recall@10	0.8914
cosine_ndcg@10	0.7878
cosine_mrr@10	0.7549
cosine_map@100	0.7596

Information Retrieval

• Dataset: dim_128
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6886
cosine_accuracy@3	0.7929
cosine_accuracy@5	0.8386
cosine_accuracy@10	0.8914
cosine_precision@1	0.6886
cosine_precision@3	0.2643
cosine_precision@5	0.1677
cosine_precision@10	0.0891
cosine_recall@1	0.6886
cosine_recall@3	0.7929
cosine_recall@5	0.8386
cosine_recall@10	0.8914
cosine_ndcg@10	0.7855
cosine_mrr@10	0.7522
cosine_map@100	0.7561

Information Retrieval

• Dataset: dim_64
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.64
cosine_accuracy@3	0.7643
cosine_accuracy@5	0.8114
cosine_accuracy@10	0.8671
cosine_precision@1	0.64
cosine_precision@3	0.2548
cosine_precision@5	0.1623
cosine_precision@10	0.0867
cosine_recall@1	0.64
cosine_recall@3	0.7643
cosine_recall@5	0.8114
cosine_recall@10	0.8671
cosine_ndcg@10	0.7492
cosine_mrr@10	0.712
cosine_map@100	0.7168

Training Details

Training Dataset

Unnamed Dataset

• Size: 6,300 training samples
• Columns: positive and anchor
• Approximate statistics based on the first 1000 samples:

  	positive	anchor
  type	string	string
  details	min: 8 tokens mean: 46.18 tokens max: 512 tokens	min: 7 tokens mean: 20.64 tokens max: 42 tokens

• Samples:

  positive	anchor
  Within the contiguous U.S., FedEx Freight offers FedEx Freight Priority, when speed is critical to meet a customer’s supply chain needs.	How does FedEx Freight accommodate rapid delivery needs?
  For purposes of our goodwill impairment evaluation, the reporting units are Family Dollar, Dollar Tree and Dollar Tree Canada.	What reporting units are used for the goodwill impairment evaluation?
  In 2024, AT&T Inc. expects a long-term rate of return of 7.75% on pension plan assets, reflecting an increase of 0.25%. This adjustment in expected returns is based on economic forecasts and changes in the asset mix.	What will AT&T Inc.'s expected long-term rate of return be on pension plan assets in 2024?

• Loss: MatryoshkaLoss with these parameters:

  {
      "loss": "MultipleNegativesRankingLoss",
      "matryoshka_dims": [
          768,
          512,
          256,
          128,
          64
      ],
      "matryoshka_weights": [
          1,
          1,
          1,
          1,
          1
      ],
      "n_dims_per_step": -1
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: epoch
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 16
• gradient_accumulation_steps: 16
• learning_rate: 2e-05
• num_train_epochs: 4
• lr_scheduler_type: cosine
• warmup_ratio: 0.1
• bf16: True
• tf32: True
• load_best_model_at_end: True
• optim: adamw_torch_fused
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: epoch
• prediction_loss_only: True
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 16
• per_gpu_train_batch_size: None
• per_gpu_eval_batch_size: None
• gradient_accumulation_steps: 16
• eval_accumulation_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: 4
• max_steps: -1
• lr_scheduler_type: cosine
• 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: True
• fp16: False
• fp16_opt_level: O1
• half_precision_backend: auto
• bf16_full_eval: False
• fp16_full_eval: False
• tf32: True
• 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: True
• 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_fused
• 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
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	dim_128_cosine_map@100	dim_256_cosine_map@100	dim_512_cosine_map@100	dim_64_cosine_map@100	dim_768_cosine_map@100
0.8122	10	1.5825	-	-	-	-	-
0.9746	12	-	0.7349	0.7502	0.7566	0.6910	0.7566
1.6244	20	0.6595	-	-	-	-	-
1.9492	24	-	0.7508	0.7583	0.7648	0.7142	0.7615
2.4365	30	0.4717	-	-	-	-	-
2.9239	36	-	0.7562	0.7616	0.7692	0.7178	0.7622
3.2487	40	0.4059	-	-	-	-	-
3.8985	48	-	0.7561	0.7596	0.7693	0.7168	0.7640

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.0.1
• Transformers: 4.41.2
• PyTorch: 2.3.1+cu121
• Accelerate: 0.32.1
• Datasets: 2.19.1
• 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",
}

MatryoshkaLoss

@misc{kusupati2024matryoshka,
    title={Matryoshka Representation Learning}, 
    author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
    year={2024},
    eprint={2205.13147},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

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}
}