You can deploy this model and start earning money today!

shivamsharma1967/
bge-base-financial-matryoshka

Name: bge-base-financial-matryoshka
Availability: InStock

Views

No views yet

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}) 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.

python

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("shivamsharma1967/bge-base-financial-matryoshka")
# Run inference
sentences = [
    'Table of Contents\nAMAZON.COM, INC.\nCONSOLIDATED STATEMENTS OF OPERATIONS\n(in millions, except per share data)\n \n \nYear Ended December 31,\n \n2015\n \n2016\n \n2017\nNet product sales\n$\n79,268 $\n94,665 $\n118,573\nNet service sales\n27,738 \n41,322 \n59,293\nTotal net sales\n107,006 \n135,987 \n177,866\nOperating expenses:\n \n \n \nCost of sales\n71,651 \n88,265 \n111,934\nFulfillment\n13,410 \n17,619 \n25,249\nMarketing\n5,254 \n7,233 \n10,069\nTechnology and content\n12,540 \n16,085 \n22,620\nGeneral and administrative\n1,747 \n2,432 \n3,674\nOther operating expense, net\n171 \n167 \n214\nTotal operating expenses\n104,773 \n131,801 \n173,760\nOperating income\n2,233 \n4,186 \n4,106\nInterest income\n50 \n100 \n202\nInterest expense\n(459) \n(484) \n(848)\nOther income (expense), net\n(256) \n90 \n346\nTotal non-operating income (expense)\n(665) \n(294) \n(300)\nIncome before income taxes\n1,568 \n3,892 \n3,806\nProvision for income taxes\n(950) \n(1,425) \n(769)\nEquity-method investment activity, net of tax\n(22) \n(96) \n(4)\nNet income\n$\n596 $\n2,371 $\n3,033\nBasic earnings per share\n$\n1.28 $\n5.01 $\n6.32\nDiluted earnings per share\n$\n1.25 $\n4.90 $\n6.15\nWeighted-average shares used in computation of earnings per share:\n \n \n \nBasic\n467 \n474 \n480\nDiluted\n477 \n484 \n493\nSee accompanying notes to consolidated financial statements.\n38\nTable of Contents\nAMAZON.COM, INC.\nCONSOLIDATED STATEMENTS OF OPERATIONS\n(in millions, except per share data)\n \n \nYear Ended December 31,\n \n2015\n \n2016\n \n2017\nNet product sales\n$\n79,268 $\n94,665 $\n118,573\nNet service sales\n27,738 \n41,322 \n59,293\nTotal net sales\n107,006 \n135,987 \n177,866\nOperating expenses:\n \n \n \nCost of sales\n71,651 \n88,265 \n111,934\nFulfillment\n13,410 \n17,619 \n25,249\nMarketing\n5,254 \n7,233 \n10,069\nTechnology and content\n12,540 \n16,085 \n22,620\nGeneral and administrative\n1,747 \n2,432 \n3,674\nOther operating expense, net\n171 \n167 \n214\nTotal operating expenses\n104,773 \n131,801 \n173,760\nOperating income\n2,233 \n4,186 \n4,106\nInterest income\n50 \n100 \n202\nInterest expense\n(459) \n(484) \n(848)\nOther income (expense), net\n(256) \n90 \n346\nTotal non-operating income (expense)\n(665) \n(294) \n(300)\nIncome before income taxes\n1,568 \n3,892 \n3,806\nProvision for income taxes\n(950) \n(1,425) \n(769)\nEquity-method investment activity, net of tax\n(22) \n(96) \n(4)\nNet income\n$\n596 $\n2,371 $\n3,033\nBasic earnings per share\n$\n1.28 $\n5.01 $\n6.32\nDiluted earnings per share\n$\n1.25 $\n4.90 $\n6.15\nWeighted-average shares used in computation of earnings per share:\n \n \n \nBasic\n467 \n474 \n480\nDiluted\n477 \n484 \n493\nSee accompanying notes to consolidated financial statements.\n38',
    "What is Amazon's year-over-year change in revenue from FY2016 to FY2017 (in units of percents and round to one decimal place)? Calculate what was asked by utilizing the line items clearly shown in the statement of income.",
    'What is the FY2018 - FY2020 3 year average of capex as a % of revenue for MGM Resorts? Answer in units of percents and round to one decimal place. Please utilize information provided primarily within the statement of cash flows and the statement of income.',
]
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

Datasets: dim_768, dim_512, dim_256, dim_128 and dim_64
Evaluated with InformationRetrievalEvaluator

Metric	dim_768	dim_512	dim_256	dim_128	dim_64
cosine_accuracy@1	0.4	0.2667	0.2	0.2	0.2667
cosine_accuracy@3	0.4667	0.4667	0.4	0.3333	0.2667
cosine_accuracy@5	0.5333	0.5333	0.4	0.4	0.3333
cosine_accuracy@10	0.6667	0.6667	0.6	0.5333	0.4667
cosine_precision@1	0.4	0.2667	0.2	0.2	0.2667
cosine_precision@3	0.1556	0.1556	0.1333	0.1111	0.0889
cosine_precision@5	0.1067	0.1067	0.08	0.08	0.0667
cosine_precision@10	0.0667	0.0667	0.06	0.0533	0.0467
cosine_recall@1	0.4	0.2667	0.2	0.2	0.2667
cosine_recall@3	0.4667	0.4667	0.4	0.3333	0.2667
cosine_recall@5	0.5333	0.5333	0.4	0.4	0.3333
cosine_recall@10	0.6667	0.6667	0.6	0.5333	0.4667
cosine_ndcg@10	0.5029	0.4537	0.374	0.346	0.3413
cosine_mrr@10	0.4541	0.3874	0.3051	0.2883	0.304
cosine_map@100	0.467	0.4024	0.3253	0.306	0.322

Training Details

Training Dataset

json

Dataset: json
Size: 135 training samples
Columns: positive and anchor
Approximate statistics based on the first 135 samples:
positive anchor
type string string
details
min: 359 tokens
mean: 508.73 tokens
max: 512 tokens
min: 11 tokens
mean: 39.7 tokens
max: 175 tokens

Samples:

positive	anchor
Twelve Months Ended June 30, 2022[object Object]Twelve Months Ended June 30, 2023[object Object]($ million)[object Object]EBITDA[object Object]EBIT[object Object]Net [object Object]Income[object Object]EPS [object Object](Diluted[object Object]US [object Object]cents)(1)[object Object]EBITDA[object Object]EBIT[object Object]Net [object Object]Income[object Object]EPS [object Object](Diluted [object Object]US [object Object]cents)(1)[object Object]Net income attributable to Amcor[object Object] [object Object]805 [object Object] [object Object]805 [object Object] [object Object]805 [object Object] [object Object]52.9 [object Object] [object Object]1,048 [object Object] [object Object]1,048 [object Object] [object Object]1,048 [object Object] [object Object]70.5 [object Object]Net income attributable to non-controlling [object Object]interests[object Object] [object Object]10 [object Object] [object Object]10 [object Object] [object Object]10 [object Object] [object Object]10 [object Object]Tax expense[object Object] [object Object]300 [object Object] [object Object]300 [object Object] [object Object]193 [object Object] [object Object]193 [object Object]Interest expense, net[object Object] [object Object]135 [object Object] [object Object]135 [object Object] [object Object]259 [object Object] [object Object]259 [object Object]Depreciation and amortization[object Object] [object Object]579 [object Object] [object Object]569 [object Object]EBITDA, EBIT, Net income and EPS[object Object] [object Object]1,829 [object Object] [object Object]1,250 [object Object] [object Object]805 [object Object] [object Object]52.9 [object Object] [object Object]2,080 [object Object] [object Object]1,510 [object Object] [object Object]1,048 [object Object] [object Object]70.5 [object Object]2019 Bemis Integration Plan[object Object] [object Object]37 [object Object] [object Object]37 [object Object] [object Object]37 [object Object] [object Object]2.5 [object Object] [object Object] [object Object] [object Object] [object Object] [object Object] [object Object] [object Object] [object Object]Net loss on disposals(2)[object Object] [object Object]10 [object Object] [object Object]10 [object Object] [object Object]10 [object Object] [object Object]0.7 [object Object] [object Object] [object Object] [object Object] [object Object] [object Object] [object Object] [object Object] [object Object]Impact of hyperinflation[object Object] [object Object]16 [object Object] [object Object]16 [object Object] [object Object]16 [object Object] [object Object]1.0 [object Object] [object Object]24 [object Object] [object Object]24 [object Object] [object Object]24 [object Object] [object Object]1.9 [object Object]Property and other losses, net(3)[object Object] [object Object]13 [object Object] [object Object]13 [object Object] [object Object]13 [object Object] [object Object]0.8 [object Object] [object Object]2 [object Object] [object Object]2 [object Object] [object Object]2 [object Object] [object Object]0.1 [object Object]Russia-Ukraine conflict impacts(4)[object Object] [object Object]200 [object Object] [object Object]200 [object Object] [object Object]200 [object Object] [object Object]13.2 [object Object] [object Object](90) [object Object](90) [object Object](90) [object Object](6.0) [object Object]Pension settlements[object Object] [object Object]8...	`What Was AMCOR's Adjusted Non GAAP EBITDA for FY 2023`
SQUARE,INC.[object Object]CONSOLIDATEDBALANCESHEETS[object Object](In thousands, except share and per share data)[object Object][object Object]December31,[object Object][object Object]2016[object Object][object Object]2015[object Object]Assets[object Object][object Object] [object Object]Currentassets:[object Object][object Object] [object Object]Cashandcashequivalents[object Object]$[object Object]452,030 $[object Object]461,329[object Object]Short-terminvestments[object Object]59,901 [object Object][object Object]Restrictedcash[object Object]22,131 [object Object]13,537[object Object]Settlementsreceivable[object Object]321,102 [object Object]142,727[object Object]Customerfundsheld[object Object]43,574 [object Object]9,446[object Object]Loansheldforsale[object Object]42,144 [object Object]604[object Object]Merchantcashadvancereceivable,net[object Object]4,212 [object Object]36,473[object Object]Othercurrentassets[object Object]56,331 [object Object]41,447[object Object]Totalcurrentassets[object Object]1,001,425 [object Object]705,563[object Object]Propertyandequipment,net[object Object]88,328 [object Object]87,222[object Object]Goodwill[object Object]57,173 [object Object]56,699[object Object]Acquiredintangibleassets,net[object Object]19,292 [object Object]26,776[object Object]Long-terminvestments[object Object]27,366 [object Object][object Object]Restrictedcash[object Object]14,584 [object Object]14,686[object Object]Otherassets[object Object]3,194 [object Object]3,826[object Object]Totalassets[object Object]$[object Object]1,211,362 $[object Object]894,772[object Object]LiabilitiesandStockholdersEquity[object Object][object Object] [object Object]Currentliabilities:[object Object][object Object] [object Object]Accountspayable[object Object]$[object Object]12,602 $[object Object]18,869[object Object]Customerspayable[object Object]388,058 [object Object]215,365[object Object]Customerfundsobligation[object Object]43,574 [object Object]9,446[object Object]Accruedtransactionlosses[object Object]20,064 [object Object]17,176[object Object]Accruedexpenses[object Object]39,543 [object Object]44,401[object Object]Othercurrentliabilities[object Object]73,623 [object Object]28,945[object Object]Totalcurrentliabilities[object Object]577,464 [object Object]33...	`Considering the data in the balance sheet, what is Block's (formerly known as Square) FY2016 working capital ratio? Define working capital ratio as total current assets divided by total current liabilities. Round your answer to two decimal places.`
Consolidated Balance Sheets [object Object]Verizon Communications Inc. and Subsidiaries [object Object](dollars in millions, except per share amounts) [object Object]At December 31,[object Object]2022[object Object]2021 [object Object]Assets [object Object]Current assets [object Object]Cash and cash equivalents[object Object]$ [object Object]2,605 [object Object]$ [object Object]2,921 [object Object]Accounts receivable[object Object] [object Object]25,332 [object Object] [object Object]24,742 [object Object]Less Allowance for credit losses[object Object] [object Object]826 [object Object] [object Object]896 [object Object]Accounts receivable, net [object Object] [object Object]24,506 [object Object] [object Object]23,846 [object Object]Inventories[object Object] [object Object]2,388 [object Object] [object Object]3,055 [object Object]Prepaid expenses and other[object Object] [object Object]8,358 [object Object] [object Object]6,906 [object Object]Total current assets[object Object] [object Object]37,857 [object Object] [object Object]36,728 [object Object]Property, plant and equipment[object Object] [object Object]307,689 [object Object] [object Object]289,897 [object Object]Less Accumulated depreciation[object Object] [object Object]200,255 [object Object] [object Object]190,201 [object Object]Property, plant and equipment, net[object Object] [object Object]107,434 [object Object] [object Object]99,696 [object Object]Investments in unconsolidated businesses[object Object] [object Object]1,071 [object Object] [object Object]1,061 [object Object]Wireless licenses[object Object] [object Object]149,796 [object Object] [object Object]147,619 [object Object]Goodwill[object Object] [object Object]28,671 [object Object] [object Object]28,603 [object Object]Other intangible assets, net[object Object] [object Object]11,461 [object Object] [object Object]11,677 [object Object]Operating lease right-of-use assets[object Object] [object Object]26,130 [object Object] [object Object]27,883 [object Object]Other assets[object Object] [object Object]17,260 [object Object] [object Object]13,329 [object Object]Total assets[object Object]$ [object Object]379,680 [object Object]$ [object Object]366,596 [object Object]Liabilities and Equity [object Object]Current liabilities [object Object]Debt maturing within o...	`Does Verizon have a reasonably healthy liquidity profile based on its quick ratio for FY 2022? If the quick ratio is not relevant to measure liquidity, please state that and explain why.`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 768, 5 512, 6 256, 7 128, 8 64 9 ], 10 "matryoshka_weights": [ 11 1, 12 1, 13 1, 14 1, 15 1 16 ], 17 "n_dims_per_step": -1 18}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: epoch
per_device_train_batch_size: 16
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: 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: 16
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
torch_empty_cache_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: 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: None
hub_always_push: False
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
include_inputs_for_metrics: False
include_for_metrics: []
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
eval_on_start: False
use_liger_kernel: False
eval_use_gather_object: False
average_tokens_across_devices: False
prompts: None
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	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	0	0.5029	0.4537	0.374	0.346	0.3413

The bold row denotes the saved checkpoint.

Framework Versions

Python: 3.11.11
Sentence Transformers: 3.4.1
Transformers: 4.48.3
PyTorch: 2.5.1+cu124
Accelerate: 1.3.0
Datasets: 3.3.2
Tokenizers: 0.21.0

Citation

BibTeX

Sentence Transformers

bibtex

@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

bibtex

@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

bibtex

@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}
}

	positive	anchor
type	string	string
details	min: 359 tokens mean: 508.73 tokens max: 512 tokens	min: 11 tokens mean: 39.7 tokens max: 175 tokens