metadata

base_model: BAAI/bge-large-en-v1.5
datasets: []
language: []
library_name: sentence-transformers
metrics:
  - pearson_cosine
  - spearman_cosine
  - pearson_manhattan
  - spearman_manhattan
  - pearson_euclidean
  - spearman_euclidean
  - pearson_dot
  - spearman_dot
  - pearson_max
  - spearman_max
pipeline_tag: sentence-similarity
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:132
  - loss:CoSENTLoss
widget:
  - source_sentence: >-
      A person shall have 3045 days after commencing business within the City to
      apply for a registration certificate.
    sentences:
      - >-
        The new transportation plan replaces the previous one approved by San
        Francisco voters in 2003. |
      - >-
        The Department of Elections is revising sections of its definitions and
        deleting a section to operate definitions for Article 12. |
      - >-
        A newly-established business shall have 3045 days after commencing
        business within the City to apply for a registration certificate, and
        the registration fee for such businesses shall be prorated based on the
        estimated gross receipts for the tax year in which the business
        commences.
  - source_sentence: >-
      The homelessness gross receipts tax is a privilege tax imposed upon
      persons engaging in business within the City for the privilege of engaging
      in a business or occupation in the City. |
    sentences:
      - >-
        The City imposes an annual Homelessness Gross Receipts Tax on businesses
        with more than $50,000,000 in total taxable gross receipts. |
      - >-
        The tax on Administrative Office Business Activities imposed by Section
        2804.9 is intended as a complementary tax to the homelessness gross
        receipts tax, and shall be considered a homelessness gross receipts tax
        for purposes of this Article 28. |
      - >-
        "The 5YPPs shall at a minimum address the following factors:
        compatibility with existing and planned land uses, and with adopted
        standards for urban design and for the provision of pedestrian
        amenities; and supportiveness of planned growth in transit-friendly
        housing, employment, and services." |
  - source_sentence: >-
      "The total worldwide compensation paid by the person and all related
      entities to the person is referred to as combined payroll." |
    sentences:
      - >-
        "A taxpayer is eligible to claim a credit against their immediately
        succeeding payments due for tax years or periods ending on or before
        December 31, 2024, of the respective tax type by applying all or part of
        an overpayment of the Homelessness Gross Receipts Tax in Article 28
        (including the homelessness administrative office tax under Section
        2804(d) of Article 28)." |
      - >-
        "Receipts from the sale of real property are exempt from the gross
        receipts tax if the Real Property Transfer Tax imposed by Article 12-C
        has been paid to the City."
      - >-
        "The total amount paid for compensation in the City by the person and by
        all related entities to the person is referred to as payroll in the
        City." |
  - source_sentence: >-
      "The gross receipts tax rates applicable to Category 6 Business Activities
      are determined based on the amount of taxable gross receipts from these
      activities." |
    sentences:
      - >-
        "The project meets the criteria outlined in Section 131051(d) of the
        Public Utilities Code."
      - >-
        For the business activity of clean technology, a tax rate of 0.175%
        (e.g. $1.75 per $1,000) applies to taxable gross receipts between $0 and
        $1,000,000 for tax years beginning on or after January 1, 2021 through
        and including 2024. |
      - >-
        "The tax rates for Category 7 Business Activities are also determined
        based on the amount of taxable gross receipts." |
  - source_sentence: >-
      "Compensation" refers to wages, salaries, commissions, bonuses, and
      property issued or transferred in exchange for services, as well as
      compensation for services to owners of pass-through entities, and any
      other form of remuneration paid to employees for services.
    sentences:
      - >-
        "Every person engaging in business within the City as an administrative
        office, as defined below, shall pay an annual administrative office tax
        measured by its total payroll expense that is attributable to the City:"
        |
      - >-
        "Remuneration" refers to any payment or reward, including but not
        limited to wages, salaries, commissions, bonuses, and property issued or
        transferred in exchange for services, as well as compensation for
        services to owners of pass-through entities, and any other form of
        compensation paid to employees for services.
      - >-
        "Construction of new Americans with Disabilities Act (ADA)-compliant
        curb ramps and related roadway work to permit ease of movement." |
model-index:
  - name: SentenceTransformer based on BAAI/bge-large-en-v1.5
    results:
      - task:
          type: semantic-similarity
          name: Semantic Similarity
        dataset:
          name: Unknown
          type: unknown
        metrics:
          - type: pearson_cosine
            value: 0.3338547038124495
            name: Pearson Cosine
          - type: spearman_cosine
            value: 0.41279297374061835
            name: Spearman Cosine
          - type: pearson_manhattan
            value: 0.3102979152053135
            name: Pearson Manhattan
          - type: spearman_manhattan
            value: 0.41673878893078603
            name: Spearman Manhattan
          - type: pearson_euclidean
            value: 0.30953937257079917
            name: Pearson Euclidean
          - type: spearman_euclidean
            value: 0.41279297374061835
            name: Spearman Euclidean
          - type: pearson_dot
            value: 0.3338548430968143
            name: Pearson Dot
          - type: spearman_dot
            value: 0.41279297374061835
            name: Spearman Dot
          - type: pearson_max
            value: 0.3338548430968143
            name: Pearson Max
          - type: spearman_max
            value: 0.41673878893078603
            name: Spearman Max

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

This is a sentence-transformers model finetuned from BAAI/bge-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.

Model Details

Model Description

Model Type: Sentence Transformer
Base model: BAAI/bge-large-en-v1.5
Maximum Sequence Length: 512 tokens
Output Dimensionality: 1024 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': 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})
  (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("Areeb-02/bge-large-en-v1.5-CosentLoss")
# Run inference
sentences = [
    '"Compensation" refers to wages, salaries, commissions, bonuses, and property issued or transferred in exchange for services, as well as compensation for services to owners of pass-through entities, and any other form of remuneration paid to employees for services.',
    '"Remuneration" refers to any payment or reward, including but not limited to wages, salaries, commissions, bonuses, and property issued or transferred in exchange for services, as well as compensation for services to owners of pass-through entities, and any other form of compensation paid to employees for services.',
    '"Every person engaging in business within the City as an administrative office, as defined below, shall pay an annual administrative office tax measured by its total payroll expense that is attributable to the City:" |',
]
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

Semantic Similarity

Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.3339
spearman_cosine	0.4128
pearson_manhattan	0.3103
spearman_manhattan	0.4167
pearson_euclidean	0.3095
spearman_euclidean	0.4128
pearson_dot	0.3339
spearman_dot	0.4128
pearson_max	0.3339
spearman_max	0.4167

Training Details

Training Dataset

Unnamed Dataset

Size: 132 training samples
Columns: sentence1, sentence2, and score
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 score
type string string float
details
min: 10 tokens
mean: 41.99 tokens
max: 126 tokens

min: 14 tokens
mean: 42.72 tokens
max: 162 tokens

min: 0.25
mean: 0.93
max: 1.0

	sentence1	sentence2	score
type	string	string	float
details	min: 10 tokens mean: 41.99 tokens max: 126 tokens	min: 14 tokens mean: 42.72 tokens max: 162 tokens	min: 0.25 mean: 0.93 max: 1.0

Samples:

sentence1	sentence2	score
`"Gross receipts as defined in Section 952.3 shall not include receipts from any sales of real property with respect to which the Real Property Transfer Tax imposed by Article 12-C has been paid to the City."`	`"Receipts from the sale of real property are exempt from the gross receipts tax if the Real Property Transfer Tax imposed by Article 12-C has been paid to the City."`	`1.0`
For tax years beginning on or after January 1, 2025, any person or combined group, except for a lessor of residential real estate, whose gross receipts within the City did not exceed $5,000,000, adjusted annually in accordance with the increase in the Consumer Price Index: All Urban Consumers for the San Francisco/Oakland/Hayward Area for All Items as reported by the United States Bureau of Labor Statistics, or any successor to that index, as of December 31 of the calendar year two years prior to the tax year, beginning with tax year 2026, and rounded to the nearest $10,000.	`For taxable years ending on or before December 31, 2024, using the rules set forth in Sections 956.1 and 956.2, in the manner directed in Sections 953.1 through 953.7, inclusive, and in Section 953.9 of this Article 12-A-1; and`	`0.95`
`"San Francisco Gross Receipts" refers to the revenue generated from sales and services within the city limits of San Francisco.`	`"Revenue generated from sales and services within the city limits of San Francisco"`	`1.0`

Loss: CoSENTLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "pairwise_cos_sim"
}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 16
per_device_eval_batch_size: 16
num_train_epochs: 5
warmup_ratio: 0.1
fp16: True

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
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: 1
eval_accumulation_steps: None
learning_rate: 5e-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: linear
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: 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: False
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
batch_sampler: batch_sampler
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	spearman_cosine
3.0	51	0.4078
5.0	45	0.4128

Framework Versions

Python: 3.10.12
Sentence Transformers: 3.0.1
Transformers: 4.42.0.dev0
PyTorch: 2.3.0+cu121
Accelerate: 0.31.0
Datasets: 2.19.2
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",
}

CoSENTLoss

@online{kexuefm-8847,
    title={CoSENT: A more efficient sentence vector scheme than Sentence-BERT},
    author={Su Jianlin},
    year={2022},
    month={Jan},
    url={https://kexue.fm/archives/8847},
}