MugheesAwan11/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("MugheesAwan11/bge-base-financial-matryoshka")
# Run inference
sentences = [
    'The net earnings margin in 2023 was 6.0%.',
    'What was the net earnings margin in 2023?',
    'What caused the slight decline in Workforce Solutions revenue in 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	0.7257
cosine_accuracy@3	0.8514
cosine_accuracy@5	0.8829
cosine_accuracy@10	0.9143
cosine_precision@1	0.7257
cosine_precision@3	0.2838
cosine_precision@5	0.1766
cosine_precision@10	0.0914
cosine_recall@1	0.7257
cosine_recall@3	0.8514
cosine_recall@5	0.8829
cosine_recall@10	0.9143
cosine_ndcg@10	0.8233
cosine_mrr@10	0.7938
cosine_map@100	0.7966

Information Retrieval

• Dataset: dim_512
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.7257
cosine_accuracy@3	0.8543
cosine_accuracy@5	0.8757
cosine_accuracy@10	0.91
cosine_precision@1	0.7257
cosine_precision@3	0.2848
cosine_precision@5	0.1751
cosine_precision@10	0.091
cosine_recall@1	0.7257
cosine_recall@3	0.8543
cosine_recall@5	0.8757
cosine_recall@10	0.91
cosine_ndcg@10	0.8215
cosine_mrr@10	0.7928
cosine_map@100	0.7959

Information Retrieval

• Dataset: dim_256
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.71
cosine_accuracy@3	0.85
cosine_accuracy@5	0.8671
cosine_accuracy@10	0.9086
cosine_precision@1	0.71
cosine_precision@3	0.2833
cosine_precision@5	0.1734
cosine_precision@10	0.0909
cosine_recall@1	0.71
cosine_recall@3	0.85
cosine_recall@5	0.8671
cosine_recall@10	0.9086
cosine_ndcg@10	0.8139
cosine_mrr@10	0.7833
cosine_map@100	0.7863

Information Retrieval

• Dataset: dim_128
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6814
cosine_accuracy@3	0.8157
cosine_accuracy@5	0.8586
cosine_accuracy@10	0.8943
cosine_precision@1	0.6814
cosine_precision@3	0.2719
cosine_precision@5	0.1717
cosine_precision@10	0.0894
cosine_recall@1	0.6814
cosine_recall@3	0.8157
cosine_recall@5	0.8586
cosine_recall@10	0.8943
cosine_ndcg@10	0.7915
cosine_mrr@10	0.7582
cosine_map@100	0.7616

Information Retrieval

• Dataset: dim_64
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.66
cosine_accuracy@3	0.78
cosine_accuracy@5	0.8071
cosine_accuracy@10	0.87
cosine_precision@1	0.66
cosine_precision@3	0.26
cosine_precision@5	0.1614
cosine_precision@10	0.087
cosine_recall@1	0.66
cosine_recall@3	0.78
cosine_recall@5	0.8071
cosine_recall@10	0.87
cosine_ndcg@10	0.7637
cosine_mrr@10	0.7301
cosine_map@100	0.7343

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: 46.61 tokens max: 289 tokens	min: 8 tokens mean: 20.58 tokens max: 45 tokens

• Samples:

  positive	anchor
  Insurance Medical Membership at December 31, 2020 for Florida includes Individual Medicare Advantage (851.3 thousand), Group Medicare Advantage (9.1 thousand), Medicare stand-alone PDP (131.9 thousand), Medicare Supplement (17.5 thousand), State-based contracts and Other (656.6 thousand), Fully-insured commercial Group (73.8 thousand), ASO (24.5 thousand), totaling 1,764.7 thousand members.	How is Florida's total insurance medical membership detailed in the data for December 31, 2023?
  For the year ended December 31, 2023, the total provision for income taxes was $836 million, which includes both current and deferred tax amounts.	What was the total provision for income taxes at the end of 2023?
  Pursuant to the IRA, under Sections 48, 48E and 25D of the Internal Revenue Code (“IRC”), standalone energy storage technology is eligible for a tax credit between 6% and 50% of qualified expenditures, regardless of the source of energy, which may be claimed by our customers for storage systems they purchase or by us for arrangements where we own the systems.	Under what sections of the Internal Revenue Code can standalone energy storage technology receive a tax credit?

• 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: 2
• 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: 2
• 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	1.4587	-	-	-	-	-
0.9746	12	-	0.7544	0.7722	0.7809	0.7118	0.7804
1.6244	20	0.6938	-	-	-	-	-
1.9492	24	-	0.7586	0.779	0.7876	0.7197	0.785
0.8122	10	0.5238	-	-	-	-	-
0.9746	12	-	0.7602	0.7815	0.7928	0.7285	0.7942
1.6244	20	0.4172	-	-	-	-	-
1.9492	24	-	0.7616	0.7863	0.7959	0.7343	0.7966

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.14
• 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}
}