dustyatx/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("dustyatx/bge-base-financial-matryoshka")
# Run inference
sentences = [
    'Details about legal proceedings are included in Part II, Item 8, "Financial Statements and Supplementary Data" of the Annual Report on Form 10-K, under the caption "Legal Proceedings".',
    'Where can details about legal proceedings be located in an Annual Report on Form 10-K?',
    'How many stores did AutoZone operate in the United States as of August 26, 2023?',
]
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	0.7071
cosine_accuracy@3	0.8414
cosine_accuracy@5	0.88
cosine_accuracy@10	0.9314
cosine_precision@1	0.7071
cosine_precision@3	0.2805
cosine_precision@5	0.176
cosine_precision@10	0.0931
cosine_recall@1	0.7071
cosine_recall@3	0.8414
cosine_recall@5	0.88
cosine_recall@10	0.9314
cosine_ndcg@10	0.8207
cosine_mrr@10	0.7853
cosine_map@100	0.7882

Information Retrieval

• Dataset: dim_512
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6957
cosine_accuracy@3	0.8386
cosine_accuracy@5	0.8757
cosine_accuracy@10	0.93
cosine_precision@1	0.6957
cosine_precision@3	0.2795
cosine_precision@5	0.1751
cosine_precision@10	0.093
cosine_recall@1	0.6957
cosine_recall@3	0.8386
cosine_recall@5	0.8757
cosine_recall@10	0.93
cosine_ndcg@10	0.8149
cosine_mrr@10	0.7781
cosine_map@100	0.781

Information Retrieval

• Dataset: dim_256
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6886
cosine_accuracy@3	0.83
cosine_accuracy@5	0.8743
cosine_accuracy@10	0.9143
cosine_precision@1	0.6886
cosine_precision@3	0.2767
cosine_precision@5	0.1749
cosine_precision@10	0.0914
cosine_recall@1	0.6886
cosine_recall@3	0.83
cosine_recall@5	0.8743
cosine_recall@10	0.9143
cosine_ndcg@10	0.8061
cosine_mrr@10	0.7711
cosine_map@100	0.7752

Information Retrieval

• Dataset: dim_128
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6771
cosine_accuracy@3	0.8214
cosine_accuracy@5	0.8614
cosine_accuracy@10	0.9143
cosine_precision@1	0.6771
cosine_precision@3	0.2738
cosine_precision@5	0.1723
cosine_precision@10	0.0914
cosine_recall@1	0.6771
cosine_recall@3	0.8214
cosine_recall@5	0.8614
cosine_recall@10	0.9143
cosine_ndcg@10	0.7979
cosine_mrr@10	0.7606
cosine_map@100	0.764

Information Retrieval

• Dataset: dim_64
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6557
cosine_accuracy@3	0.7871
cosine_accuracy@5	0.8271
cosine_accuracy@10	0.8714
cosine_precision@1	0.6557
cosine_precision@3	0.2624
cosine_precision@5	0.1654
cosine_precision@10	0.0871
cosine_recall@1	0.6557
cosine_recall@3	0.7871
cosine_recall@5	0.8271
cosine_recall@10	0.8714
cosine_ndcg@10	0.7664
cosine_mrr@10	0.7327
cosine_map@100	0.7376

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: 45.94 tokens max: 512 tokens	min: 7 tokens mean: 20.7 tokens max: 42 tokens

• Samples:

  positive	anchor
  The company must continuously strengthen its capabilities in marketing and innovation to compete in a digital environment and maintain brand loyalty and marketallability. In addition, it is increasing its investments in e-commerce to support retail and meal delivery services, offering more package sizes that are fit-for-purpose for online sales and shifting more consumer and trade promotions to digital.	What strategies is the company employing to enhance its competitiveness in a digital environment?
  Fedflowing expanded or relocated its hub and linehaul network, FedEx Ground also introduced new safety technologies, set new driver standards, and made operational enhancements for safer handling of heavy items.	What specific changes has FedEx Ground made for vehicle and driver safety?
  The debt financing, which is being provided by a syndicate of Chinese financial institutions, contains certain covenants and a maximum borrowing limit of ¥29.7 billion RMB (approximately $4.2 billion).	What is the maximum borrowing limit of the debt financing provided by the syndicate of Chinese financial institutions for Universal Beijing Resort?

• 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
• 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: 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
• eval_on_start: False
• eval_use_gather_object: 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	1.5212	-	-	-	-	-
0.9746	12	-	0.7439	0.7556	0.7670	0.7142	0.7717
1.6244	20	0.6418	-	-	-	-	-
1.9492	24	-	0.7592	0.7743	0.7787	0.7331	0.7839
2.4365	30	0.4411	-	-	-	-	-
2.9239	36	-	0.7623	0.7757	0.7816	0.7365	0.7902
3.2487	40	0.3917	-	-	-	-	-
3.8985	48	-	0.764	0.7752	0.781	0.7376	0.7882

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.11.9
• Sentence Transformers: 3.0.1
• Transformers: 4.44.2
• PyTorch: 2.4.0+cu121
• Accelerate: 0.33.0
• Datasets: 2.21.0
• 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",
}

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}
}