Yuki20/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 tokens
• 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}) 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("Yuki20/bge-base-financial-matryoshka")
# Run inference
sentences = [
    "As of December 31, 2023, the fair value and amortized cost, net of valuation allowance, for the Republic of Korea's government securities were $1,784 million and $1,723 million respectively.",
    "What was the fair value and amortized cost, net of valuation allowance, for the Republic of Korea's government securities as of December 31, 2023?",
    'How does the company advance autonomous vehicle technology?',
]
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	0.6871
cosine_accuracy@3	0.8286
cosine_accuracy@5	0.8571
cosine_accuracy@10	0.9071
cosine_precision@1	0.6871
cosine_precision@3	0.2762
cosine_precision@5	0.1714
cosine_precision@10	0.0907
cosine_recall@1	0.6871
cosine_recall@3	0.8286
cosine_recall@5	0.8571
cosine_recall@10	0.9071
cosine_ndcg@10	0.7982
cosine_mrr@10	0.7633
cosine_map@100	0.767

Information Retrieval

• Dataset: dim_512
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.69
cosine_accuracy@3	0.8171
cosine_accuracy@5	0.8543
cosine_accuracy@10	0.9043
cosine_precision@1	0.69
cosine_precision@3	0.2724
cosine_precision@5	0.1709
cosine_precision@10	0.0904
cosine_recall@1	0.69
cosine_recall@3	0.8171
cosine_recall@5	0.8543
cosine_recall@10	0.9043
cosine_ndcg@10	0.7977
cosine_mrr@10	0.7636
cosine_map@100	0.7675

Information Retrieval

• Dataset: dim_256
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6857
cosine_accuracy@3	0.8143
cosine_accuracy@5	0.8514
cosine_accuracy@10	0.8957
cosine_precision@1	0.6857
cosine_precision@3	0.2714
cosine_precision@5	0.1703
cosine_precision@10	0.0896
cosine_recall@1	0.6857
cosine_recall@3	0.8143
cosine_recall@5	0.8514
cosine_recall@10	0.8957
cosine_ndcg@10	0.7916
cosine_mrr@10	0.7582
cosine_map@100	0.7624

Information Retrieval

• Dataset: dim_128
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6757
cosine_accuracy@3	0.8
cosine_accuracy@5	0.8414
cosine_accuracy@10	0.8886
cosine_precision@1	0.6757
cosine_precision@3	0.2667
cosine_precision@5	0.1683
cosine_precision@10	0.0889
cosine_recall@1	0.6757
cosine_recall@3	0.8
cosine_recall@5	0.8414
cosine_recall@10	0.8886
cosine_ndcg@10	0.782
cosine_mrr@10	0.7478
cosine_map@100	0.7524

Information Retrieval

• Dataset: dim_64
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6414
cosine_accuracy@3	0.7657
cosine_accuracy@5	0.7957
cosine_accuracy@10	0.8586
cosine_precision@1	0.6414
cosine_precision@3	0.2552
cosine_precision@5	0.1591
cosine_precision@10	0.0859
cosine_recall@1	0.6414
cosine_recall@3	0.7657
cosine_recall@5	0.7957
cosine_recall@10	0.8586
cosine_ndcg@10	0.748
cosine_mrr@10	0.7129
cosine_map@100	0.7185

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: 6 tokens mean: 45.58 tokens max: 289 tokens	min: 9 tokens mean: 20.34 tokens max: 41 tokens

• Samples:

  positive	anchor
  Billed business grew significantly over the past two years, increasing from $228.2 billion in 2021 to $281.6 billion in 2022, and reaching $329.5 billion in 2023.	How did billed business figures change from 2021 to 2023 as stated in the text?
  The Federal Reserve may limit an FHC’s ability to conduct permissible activities if it or any of its depository institution subsidiaries fails to maintain a well-capitalized and well-managed status. If non-compliant after 180 days, the Federal Reserve may require the FHC to divest its depository institution subsidiaries or cease all FHC Activities.	What happens if an FHC does not meet the Federal Reserve's eligibility requirements?
  For the fiscal year ending January 28, 2023, the basic net income per share was calculated to be $7.24, based on the net income and weighted average number of shares outstanding.	What was the basic net income per share in the fiscal year ending January 28, 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
• fp16: True
• tf32: False
• 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: False
• fp16: True
• fp16_opt_level: O1
• half_precision_backend: auto
• bf16_full_eval: False
• fp16_full_eval: False
• tf32: False
• 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_768_cosine_map@100	dim_512_cosine_map@100	dim_256_cosine_map@100	dim_128_cosine_map@100	dim_64_cosine_map@100
0.8122	10	1.588	-	-	-	-	-
0.9746	12	-	0.7593	0.7550	0.7472	0.7347	0.6970
1.6244	20	0.7059	-	-	-	-	-
1.9492	24	-	0.7623	0.7652	0.7559	0.7517	0.7127
2.4365	30	0.4826	-	-	-	-	-
2.9239	36	-	0.7675	0.7683	0.7603	0.7512	0.7166
3.2487	40	0.3992	-	-	-	-	-
3.8985	48	-	0.767	0.7675	0.7624	0.7524	0.7185

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.2.0
• Transformers: 4.41.2
• PyTorch: 2.1.2+cu121
• Accelerate: 0.34.2
• 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}
}