metadata

library_name: sentence-transformers
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:6300
  - loss:MatryoshkaLoss
  - loss:MultipleNegativesRankingLoss
base_model: BAAI/bge-base-en-v1.5
metrics:
  - cosine_accuracy@1
  - cosine_accuracy@3
  - cosine_accuracy@5
  - cosine_accuracy@10
  - cosine_precision@1
  - cosine_precision@3
  - cosine_precision@5
  - cosine_precision@10
  - cosine_recall@1
  - cosine_recall@3
  - cosine_recall@5
  - cosine_recall@10
  - cosine_ndcg@10
  - cosine_mrr@10
  - cosine_map@100
widget:
  - source_sentence: >-
      The Financial Statements and Supplementary Data are listed under 'Item 15.
      Exhibits and Financial Statement Schedules' as part of this Report.
    sentences:
      - >-
        Under ASO contracts, who retains the risk of financing health benefits
        costs?
      - >-
        Under which item are the Financial Statements and Supplementary Data
        listed in the report?
      - >-
        How much did interest income increase for Enphase Energy in the year
        ended December 31, 2023 compared to 2022?
  - source_sentence: >-
      The company is involved in various legal actions in the ordinary course of
      business, and some of these may be styled as class-action lawsuits.
    sentences:
      - How many Dollar Tree Plus stores were there as of January 28, 2023?
      - >-
        Are there any class-action lawsuits among the legal actions faced by the
        company?
      - What are the components referred to in Item 8 of financial documents?
  - source_sentence: >-
      In 2021, the net cash provided by operating activities was $3,264 million
      and it increased to $6,464 million by 2023.
    sentences:
      - >-
        What was the net increase in cash provided by operating activities from
        2021 to 2023?
      - >-
        What specific competitive advantages does IBM leverage in the hybrid
        cloud infrastructure market?
      - >-
        How many unvested restricted stock awards were there as of December 31,
        2022, and what was the weighted-average grant price at that time?
  - source_sentence: >-
      We have assets for foreign net operating losses of $133.5 million, with
      various expiration dates (and in some cases no expiration date), subject
      to a valuation stand
    sentences:
      - >-
        What page in IBM’s 2023 Form 10-K is reserved for the Financial
        Statement Schedule?
      - >-
        What were the primary sources and uses of cash that contributed to the
        $7.8 billion increase in cash and cash equivalents during 2023?
      - >-
        What is the total value of foreign net operating losses reported, and
        what is the valuation allowance percentage applied to them?
  - source_sentence: >-
      The Company uses cash flow hedges to minimize the variability in cash
      flows of assets or liabilities or forecasted transactions caused by
      fluctuations in foreign currency exchange rates, commodity prices or
      interest. The changes in the fair values of derivatives designated as cash
      flow hedges are recorded in AOCI and are reclassified into the line item
      in our consolidated statement of income in which the hedged items are
      recorded in the same period the hedged items affect earnings.
    sentences:
      - >-
        What financial instruments does the Company use to minimize the
        variability in cash flows due to fluctuations in foreign currency
        exchange rates, commodity prices, or interest rates?
      - >-
        Why is the Asia Pacific reporting unit considered at risk of future
        goodwill impairment?
      - >-
        What constituted the majority of the cost of revenues in the discussed
        financial year?
pipeline_tag: sentence-similarity
model-index:
  - name: SentenceTransformer based on BAAI/bge-base-en-v1.5
    results:
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 512
          type: dim_512
        metrics:
          - type: cosine_accuracy@1
            value: 0.7285714285714285
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.83
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.8685714285714285
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.9128571428571428
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.7285714285714285
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.27666666666666667
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.1737142857142857
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.09128571428571428
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.7285714285714285
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.83
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.8685714285714285
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.9128571428571428
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.8196512798721632
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.7899614512471653
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.7931894941486222
            name: Cosine Map@100

SentenceTransformer based on BAAI/bge-base-en-v1.5

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

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("ppuva1/bge-base-financial-matryoshka-2")
# Run inference
sentences = [
    'The Company uses cash flow hedges to minimize the variability in cash flows of assets or liabilities or forecasted transactions caused by fluctuations in foreign currency exchange rates, commodity prices or interest. The changes in the fair values of derivatives designated as cash flow hedges are recorded in AOCI and are reclassified into the line item in our consolidated statement of income in which the hedged items are recorded in the same period the hedged items affect earnings.',
    'What financial instruments does the Company use to minimize the variability in cash flows due to fluctuations in foreign currency exchange rates, commodity prices, or interest rates?',
    'Why is the Asia Pacific reporting unit considered at risk of future goodwill impairment?',
]
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_512
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.7286
cosine_accuracy@3	0.83
cosine_accuracy@5	0.8686
cosine_accuracy@10	0.9129
cosine_precision@1	0.7286
cosine_precision@3	0.2767
cosine_precision@5	0.1737
cosine_precision@10	0.0913
cosine_recall@1	0.7286
cosine_recall@3	0.83
cosine_recall@5	0.8686
cosine_recall@10	0.9129
cosine_ndcg@10	0.8197
cosine_mrr@10	0.79
cosine_map@100	0.7932

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: 4 tokens
mean: 46.01 tokens
max: 205 tokens

min: 7 tokens
mean: 20.43 tokens
max: 46 tokens

	positive	anchor
type	string	string
details	min: 4 tokens mean: 46.01 tokens max: 205 tokens	min: 7 tokens mean: 20.43 tokens max: 46 tokens

Samples:

positive	anchor
`The company's SEC filings are available to the public over the internet at the SEC's website at www.sec.gov. The SEC filings are also available free of charge on the company's website at ir.hilton.com as soon as reasonably practicable after they are filed with or furnished to the SEC.`	`Where can public access the company's SEC filings?`
`Garmin’s operations are subject to various environmental laws, including laws addressing air and water pollution and management of hazardous substances and wastes.`	`What aspects of Garmin's business are subject to environmental laws?`
`Adjusted EBITDA does not reflect certain litigation expenses, consisting of legal settlements and related fees for specific proceedings that we have determined arise outside of the ordinary course of business.`	`How does Adjusted EBITDA treat expenses related to litigation?`

Loss: MatryoshkaLoss with these parameters:

{
    "loss": "MultipleNegativesRankingLoss",
    "matryoshka_dims": [
        512
    ],
    "matryoshka_weights": [
        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: 5
lr_scheduler_type: cosine
warmup_ratio: 0.1
load_best_model_at_end: True
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
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: 5
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: False
fp16_opt_level: O1
half_precision_backend: auto
bf16_full_eval: False
fp16_full_eval: False
tf32: None
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
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
eval_on_start: False
use_liger_kernel: False
eval_use_gather_object: False
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	dim_512_cosine_map@100
0	0	0.7260
0.9746	12	0.7815
1.9492	24	0.7872
2.9239	36	0.7899
3.9797	49	0.7926
4.8731	60	0.7932

The bold row denotes the saved checkpoint.

Framework Versions

Python: 3.11.8
Sentence Transformers: 3.1.1
Transformers: 4.45.2
PyTorch: 2.4.1
Accelerate: 0.34.2
Datasets: 3.2.0
Tokenizers: 0.20.0

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}
}