financial-rag-matryoshka

Model finetuned for financial use-cases from Alibaba-NLP/gte-large-en-v1.5. It maps sentences & paragraphs to a 1024-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

This model strives to excel tremendously in Financial Document Retrieval Tasks, concurrently preserving a maximum level of generalized performance.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: Alibaba-NLP/gte-large-en-v1.5
• Maximum Sequence Length: 8192 tokens
• Output Dimensionality: 1024 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': 8192, 'do_lower_case': False}) with Transformer model: NewModel 
  (1): Pooling({'word_embedding_dimension': 1024, '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})
)

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("rbhatia46/gte-large-en-v1.5-financial-rag-matryoshka")
# Run inference
sentences = [
    'JP Morgan reported total deposits of $2.6 trillion in the year ending December 31, 2023.',
    "What were JP Morgan's total deposits in 2023?",
    'What is the primary source of revenue for the software company, Microsoft?',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Evaluation

Metrics

Information Retrieval

• Dataset: dim_1024
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.88
cosine_accuracy@3	0.96
cosine_accuracy@5	0.9867
cosine_accuracy@10	0.9956
cosine_precision@1	0.88
cosine_precision@3	0.32
cosine_precision@5	0.1973
cosine_precision@10	0.0996
cosine_recall@1	0.88
cosine_recall@3	0.96
cosine_recall@5	0.9867
cosine_recall@10	0.9956
cosine_ndcg@10	0.9427
cosine_mrr@10	0.9252
cosine_map@100	0.9254

Information Retrieval

• Dataset: dim_768
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.88
cosine_accuracy@3	0.96
cosine_accuracy@5	0.9867
cosine_accuracy@10	0.9911
cosine_precision@1	0.88
cosine_precision@3	0.32
cosine_precision@5	0.1973
cosine_precision@10	0.0991
cosine_recall@1	0.88
cosine_recall@3	0.96
cosine_recall@5	0.9867
cosine_recall@10	0.9911
cosine_ndcg@10	0.9408
cosine_mrr@10	0.924
cosine_map@100	0.9245

Information Retrieval

• Dataset: dim_512
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.8711
cosine_accuracy@3	0.96
cosine_accuracy@5	0.9867
cosine_accuracy@10	0.9911
cosine_precision@1	0.8711
cosine_precision@3	0.32
cosine_precision@5	0.1973
cosine_precision@10	0.0991
cosine_recall@1	0.8711
cosine_recall@3	0.96
cosine_recall@5	0.9867
cosine_recall@10	0.9911
cosine_ndcg@10	0.9381
cosine_mrr@10	0.9203
cosine_map@100	0.9207

Information Retrieval

• Dataset: dim_256
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.8756
cosine_accuracy@3	0.96
cosine_accuracy@5	0.9867
cosine_accuracy@10	0.9911
cosine_precision@1	0.8756
cosine_precision@3	0.32
cosine_precision@5	0.1973
cosine_precision@10	0.0991
cosine_recall@1	0.8756
cosine_recall@3	0.96
cosine_recall@5	0.9867
cosine_recall@10	0.9911
cosine_ndcg@10	0.9396
cosine_mrr@10	0.9223
cosine_map@100	0.9228

Information Retrieval

• Dataset: dim_128
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.8667
cosine_accuracy@3	0.9556
cosine_accuracy@5	0.9867
cosine_accuracy@10	0.9911
cosine_precision@1	0.8667
cosine_precision@3	0.3185
cosine_precision@5	0.1973
cosine_precision@10	0.0991
cosine_recall@1	0.8667
cosine_recall@3	0.9556
cosine_recall@5	0.9867
cosine_recall@10	0.9911
cosine_ndcg@10	0.9346
cosine_mrr@10	0.9157
cosine_map@100	0.916

Information Retrieval

• Dataset: dim_64
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.8311
cosine_accuracy@3	0.96
cosine_accuracy@5	0.9733
cosine_accuracy@10	0.9911
cosine_precision@1	0.8311
cosine_precision@3	0.32
cosine_precision@5	0.1947
cosine_precision@10	0.0991
cosine_recall@1	0.8311
cosine_recall@3	0.96
cosine_recall@5	0.9733
cosine_recall@10	0.9911
cosine_ndcg@10	0.9208
cosine_mrr@10	0.8972
cosine_map@100	0.8975

Training Details

Training Dataset

Unnamed Dataset

• Size: 4,275 training samples
• Columns: positive and anchor
• Approximate statistics based on the first 1000 samples:

  	positive	anchor
  type	string	string
  details	min: 15 tokens mean: 44.74 tokens max: 114 tokens	min: 9 tokens mean: 18.12 tokens max: 32 tokens

• Samples:

  positive	anchor
  At the end of fiscal year 2023, Exxon Mobil reported a debt-to-equity ratio of 0.32, implying that the company used more equity than debt in its capital structure.	What was the debt-to-equity ratio for Exxon Mobil at the end of fiscal year 2023?
  Amazon Web Services (AWS) generated $12.7 billion in net sales in the fourth quarter of 2020, up 28% from the same period of the previous year. It accounted for about 10% of Amazon’s total net sales for the quarter.	How did Amazon's AWS segment perform in the fourth quarter of 2020?
  JPMorgan Chase generates revenues by providing a wide range of banking and financial services. These include investment banking (M&As, advisory), consumer and community banking (home mortgages, auto loans), commercial banking, and asset and wealth management.	What are the key revenue sources for JPMorgan Chase?

• Loss: MatryoshkaLoss with these parameters:

  {
      "loss": "MultipleNegativesRankingLoss",
      "matryoshka_dims": [
          1024,
          768,
          512,
          256,
          128,
          64
      ],
      "matryoshka_weights": [
          1,
          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: 10
• 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: 10
• 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_1024_cosine_map@100	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
0.9552	8	-	0.9090	0.8848	0.8992	0.9052	0.8775	0.9030
1.1940	10	0.4749	-	-	-	-	-	-
1.9104	16	-	0.9170	0.9095	0.9109	0.9201	0.8961	0.9212
2.3881	20	0.0862	-	-	-	-	-	-
2.9851	25	-	0.9190	0.9071	0.9160	0.9278	0.8998	0.9234
3.5821	30	0.0315	-	-	-	-	-	-
3.9403	33	-	0.9183	0.9053	0.9122	0.9287	0.8998	0.9183
4.7761	40	0.0184	-	-	-	-	-	-
4.8955	41	-	0.9225	0.9125	0.9164	0.9260	0.8985	0.9220
5.9701	50	0.0135	0.9268	0.9132	0.9208	0.9257	0.8961	0.9271
6.9254	58	-	0.9254	0.9158	0.9202	0.9212	0.8938	0.9213
7.1642	60	0.0123	-	-	-	-	-	-
8.0	67	-	0.9253	0.916	0.9228	0.9207	0.8972	0.9243
8.3582	70	0.01	-	-	-	-	-	-
8.9552	75	-	0.9254	0.9160	0.9213	0.9207	0.9005	0.9245
9.5522	80	0.0088	0.9254	0.9160	0.9228	0.9207	0.8975	0.9245

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.6
• Sentence Transformers: 3.0.1
• Transformers: 4.41.2
• PyTorch: 2.1.2+cu121
• Accelerate: 0.32.1
• 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}
}