uonyeka/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("uonyeka/bge-base-financial-matryoshka")
# Run inference
sentences = [
    'The return on equity for 2023 was 27.0%.',
    'What was the return on equity for 2023?',
    'What was the total net property and equipment as of December 31, 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	0.7029
cosine_accuracy@3	0.84
cosine_accuracy@5	0.8829
cosine_accuracy@10	0.9214
cosine_precision@1	0.7029
cosine_precision@3	0.28
cosine_precision@5	0.1766
cosine_precision@10	0.0921
cosine_recall@1	0.7029
cosine_recall@3	0.84
cosine_recall@5	0.8829
cosine_recall@10	0.9214
cosine_ndcg@10	0.8149
cosine_mrr@10	0.7804
cosine_map@100	0.7839

Information Retrieval

• Dataset: dim_512
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.7029
cosine_accuracy@3	0.8386
cosine_accuracy@5	0.8871
cosine_accuracy@10	0.9257
cosine_precision@1	0.7029
cosine_precision@3	0.2795
cosine_precision@5	0.1774
cosine_precision@10	0.0926
cosine_recall@1	0.7029
cosine_recall@3	0.8386
cosine_recall@5	0.8871
cosine_recall@10	0.9257
cosine_ndcg@10	0.8153
cosine_mrr@10	0.7797
cosine_map@100	0.783

Information Retrieval

• Dataset: dim_256
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.7057
cosine_accuracy@3	0.8329
cosine_accuracy@5	0.8729
cosine_accuracy@10	0.9114
cosine_precision@1	0.7057
cosine_precision@3	0.2776
cosine_precision@5	0.1746
cosine_precision@10	0.0911
cosine_recall@1	0.7057
cosine_recall@3	0.8329
cosine_recall@5	0.8729
cosine_recall@10	0.9114
cosine_ndcg@10	0.8097
cosine_mrr@10	0.777
cosine_map@100	0.7811

Information Retrieval

• Dataset: dim_128
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.68
cosine_accuracy@3	0.8286
cosine_accuracy@5	0.8671
cosine_accuracy@10	0.91
cosine_precision@1	0.68
cosine_precision@3	0.2762
cosine_precision@5	0.1734
cosine_precision@10	0.091
cosine_recall@1	0.68
cosine_recall@3	0.8286
cosine_recall@5	0.8671
cosine_recall@10	0.91
cosine_ndcg@10	0.7972
cosine_mrr@10	0.7608
cosine_map@100	0.7644

Information Retrieval

• Dataset: dim_64
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6614
cosine_accuracy@3	0.7943
cosine_accuracy@5	0.83
cosine_accuracy@10	0.8757
cosine_precision@1	0.6614
cosine_precision@3	0.2648
cosine_precision@5	0.166
cosine_precision@10	0.0876
cosine_recall@1	0.6614
cosine_recall@3	0.7943
cosine_recall@5	0.83
cosine_recall@10	0.8757
cosine_ndcg@10	0.7702
cosine_mrr@10	0.7363
cosine_map@100	0.7414

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: 6 tokens mean: 45.2 tokens max: 439 tokens	min: 7 tokens mean: 20.41 tokens max: 43 tokens

• Samples:

  positive	anchor
  The cash equities rate per contract (per 100 shares) for NYSE increased by 6%, from $0.045 in 2022 to $0.048 in 2023.	What was the change in the rate per contract for NYSE cash equities from 2022 to 2023?
  Item 3 specifies that the information regarding Legal Proceedings is sourced from Note 19 of the Notes to Consolidated Financial Statements included in Item 8.	What is the content source for the information requested by Item 3 concerning Legal Proceedings?
  North America's operating income for the fiscal year ended October 1, 2023, was $5,495.7 million, up from $4,486.5 million in fiscal 2022.	What was the increase in North America's operating income from fiscal 2022 to fiscal 2023?

• 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	1.575	-	-	-	-	-
0.9746	12	-	0.7437	0.7623	0.7682	0.7114	0.7652
1.6244	20	0.697	-	-	-	-	-
1.9492	24	-	0.7619	0.7760	0.7824	0.7346	0.7826
2.4365	30	0.4724	-	-	-	-	-
2.9239	36	-	0.7639	0.7808	0.7831	0.7398	0.7834
3.2487	40	0.3999	-	-	-	-	-
3.8985	48	-	0.7644	0.7811	0.783	0.7414	0.7839

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.0.0
• Transformers: 4.41.2
• PyTorch: 2.1.2+cu121
• Accelerate: 0.30.1
• 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}
}