deter3/bge-base-financial-matryoshka

BGE base Financial Matryoshka

This is a sentence-transformers model finetuned from BAAI/bge-base-en-v1.5 on the json dataset. 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 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • json
• 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, 'architecture': '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("deter3/bge-base-financial-matryoshka")
# Run inference
sentences = [
    'Generated cash flows from operations of $4.5 billion.',
    'How much did cash flows from operations amount to in 2022?',
    'What was the overall turnover rate at the company in fiscal year 2023?',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.7518, 0.2425],
#         [0.7518, 1.0000, 0.2768],
#         [0.2425, 0.2768, 1.0000]])

Evaluation

Metrics

Information Retrieval

• Dataset: dim_768
• Evaluated with InformationRetrievalEvaluator with these parameters:

  {
      "truncate_dim": 768
  }
  

Metric	Value
cosine_accuracy@1	0.7014
cosine_accuracy@3	0.8243
cosine_accuracy@5	0.8671
cosine_accuracy@10	0.9071
cosine_precision@1	0.7014
cosine_precision@3	0.2748
cosine_precision@5	0.1734
cosine_precision@10	0.0907
cosine_recall@1	0.7014
cosine_recall@3	0.8243
cosine_recall@5	0.8671
cosine_recall@10	0.9071
cosine_ndcg@10	0.8053
cosine_mrr@10	0.7727
cosine_map@100	0.7764

Information Retrieval

• Dataset: dim_512
• Evaluated with InformationRetrievalEvaluator with these parameters:

  {
      "truncate_dim": 512
  }
  

Metric	Value
cosine_accuracy@1	0.7114
cosine_accuracy@3	0.8243
cosine_accuracy@5	0.8643
cosine_accuracy@10	0.9086
cosine_precision@1	0.7114
cosine_precision@3	0.2748
cosine_precision@5	0.1729
cosine_precision@10	0.0909
cosine_recall@1	0.7114
cosine_recall@3	0.8243
cosine_recall@5	0.8643
cosine_recall@10	0.9086
cosine_ndcg@10	0.8099
cosine_mrr@10	0.7784
cosine_map@100	0.782

Information Retrieval

• Dataset: dim_256
• Evaluated with InformationRetrievalEvaluator with these parameters:

  {
      "truncate_dim": 256
  }
  

Metric	Value
cosine_accuracy@1	0.7014
cosine_accuracy@3	0.8243
cosine_accuracy@5	0.8557
cosine_accuracy@10	0.8914
cosine_precision@1	0.7014
cosine_precision@3	0.2748
cosine_precision@5	0.1711
cosine_precision@10	0.0891
cosine_recall@1	0.7014
cosine_recall@3	0.8243
cosine_recall@5	0.8557
cosine_recall@10	0.8914
cosine_ndcg@10	0.8009
cosine_mrr@10	0.7715
cosine_map@100	0.7759

Information Retrieval

• Dataset: dim_128
• Evaluated with InformationRetrievalEvaluator with these parameters:

  {
      "truncate_dim": 128
  }
  

Metric	Value
cosine_accuracy@1	0.6829
cosine_accuracy@3	0.8129
cosine_accuracy@5	0.8486
cosine_accuracy@10	0.8914
cosine_precision@1	0.6829
cosine_precision@3	0.271
cosine_precision@5	0.1697
cosine_precision@10	0.0891
cosine_recall@1	0.6829
cosine_recall@3	0.8129
cosine_recall@5	0.8486
cosine_recall@10	0.8914
cosine_ndcg@10	0.7894
cosine_mrr@10	0.7566
cosine_map@100	0.7607

Information Retrieval

• Dataset: dim_64
• Evaluated with InformationRetrievalEvaluator with these parameters:

  {
      "truncate_dim": 64
  }
  

Metric	Value
cosine_accuracy@1	0.6614
cosine_accuracy@3	0.7957
cosine_accuracy@5	0.8286
cosine_accuracy@10	0.8771
cosine_precision@1	0.6614
cosine_precision@3	0.2652
cosine_precision@5	0.1657
cosine_precision@10	0.0877
cosine_recall@1	0.6614
cosine_recall@3	0.7957
cosine_recall@5	0.8286
cosine_recall@10	0.8771
cosine_ndcg@10	0.7707
cosine_mrr@10	0.7366
cosine_map@100	0.7409

Training Details

Training Dataset

json

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

  	positive	anchor
  type	string	string
  details	min: 13 tokens mean: 45.95 tokens max: 248 tokens	min: 8 tokens mean: 20.43 tokens max: 41 tokens

• Samples:

  positive	anchor
  The Company's nominal par value per share was slightly reduced to USD $0.10, reflecting in the share capital, as of December 30, 2023.	What was the nominal par value per share of Garmin Ltd. in U.S. dollars as of December 30, 2023?
  Over the last several years, the number and potential significance of the litigation and investigations involving the company have increased, and there can be no assurance that this trend will not continue.	How has the litigation and investigation landscape changed for the company over recent years?
  As of January 31, 2023, assets located outside the Americas were 15 percent of total assets.	What percentage of Salesforce's total assets were located outside the Americas as of January 31, 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
• prompts: None
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional
• router_mapping: {}
• learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss	dim_768_cosine_ndcg@10	dim_512_cosine_ndcg@10	dim_256_cosine_ndcg@10	dim_128_cosine_ndcg@10	dim_64_cosine_ndcg@10
0.8122	10	1.5429	-	-	-	-	-
0.9746	12	-	0.7915	0.7927	0.7820	0.7722	0.7396
1.6244	20	0.6772	-	-	-	-	-
1.9492	24	-	0.8019	0.8041	0.7971	0.7835	0.7625
2.4365	30	0.5496	-	-	-	-	-
2.9239	36	-	0.8048	0.8070	0.8007	0.7879	0.7690
3.2487	40	0.4528	-	-	-	-	-
3.8985	48	-	0.8053	0.8099	0.8009	0.7894	0.7707

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 5.0.0
• Transformers: 4.41.2
• PyTorch: 2.1.2+cu121
• Accelerate: 1.8.1
• 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}
}