metadata

language:
  - en
license: apache-2.0
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:843
  - loss:MatryoshkaLoss
  - loss:MultipleNegativesRankingLoss
base_model: BAAI/bge-base-en-v1.5
widget:
  - source_sentence: >-
      (1) No person shall make attempt to commit an offence. Even if it is
      impossible for an offence to be committed for which attempt is made,
      attempt shall be considered to have been committed. Except as otherwise
      provided elsewhere in this Act, a person who attempts, or causes attempt,
      to commit an offence shall be punished with one half of the punishment
      specified for such offence. .
    sentences:
      - How is the punishment for an attempt determined?
      - What are the different types of guarantees?
      - >-
        What are the specific types of crimes that are considered 'strict
        liability'?
  - source_sentence: >-
      : (1) No person shall commit, or cause to be committed, cheating. (2) For
      the purposes of sub-section (1), a person who dishonestly causes any kind
      of loss, damage or injury to another person whom he or she makes believe
      in some matter or to any other person or obtains any benefit for him or
      her or any one else by omitting to do as per such belief or by inducement,
      fraudulent, dishonest or otherwise deceptive act or preventing such other
      person from doing any act shall be considered to commit cheating.
    sentences:
      - How is 'fraudulent concealment' defined?
      - >-
        What are the terms and restrictions that must be followed when producing
        explosives under a license?
      - >-
        What is the process for determining the appropriate penalty for a
        cheating offense?
  - source_sentence: >-
      (1) No person shall restraint or otherwise obstruct or hinder a person
      who, upon knowing that an offence has been committed or is about to be
      committed, intends to give information or notice about such offence to the
      police or competent authority. imprisonment for a term not exceeding two
      years or a fine not exceeding twenty thousand rupees or both the
      sentences.
    sentences:
      - What actions constitute 'restraint, obstruction, or hindrance'?
      - What are the consequences of engaging in such conduct?
      - >-
        What are the different categories of victims, and how do the penalties
        vary based on their age?
  - source_sentence: >-
      This law prohibits the creation, use, possession, or sale of inaccurate
      weighing, measuring, or quality-standard instruments.  It also prohibits
      tampering with seals or marks on these instruments, or manipulating their
      accuracy.  Violations carry a penalty of up to three years imprisonment
      and a fine.  Instruments and tools used in the offense are subject to
      forfeiture.
    sentences:
      - What are the penalties for using banned currency?
      - What is the time frame for reporting an offense under this law?
      - When does this law come into effect?
  - source_sentence: >-
      This section lists factors that decrease the seriousness of a crime. 
      These include age (under 18 or over 75), lack of intent, provocation by
      the victim, retaliation for a serious offense, confession and remorse,
      surrender to authorities, compensation to the victim, diminished capacity,
      insignificant harm, assistance in the judicial process, confession with a
      promise of no future crime, and crimes committed under duress.
    sentences:
      - What constitutes "lack of intent" in this context?
      - >-
        What is the difference between an attempt and the actual commission of a
        crime?
      - >-
        What are the exceptions to the prohibition on property transactions in
        marriage?
pipeline_tag: sentence-similarity
library_name: sentence-transformers
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
model-index:
  - name: BGE base Financial Matryoshka
    results:
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 128
          type: dim_128
        metrics:
          - type: cosine_accuracy@1
            value: 0.13744075829383887
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.4312796208530806
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.5450236966824644
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.6445497630331753
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.13744075829383887
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.14375987361769352
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.10900473933649289
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.06445497630331752
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.13744075829383887
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.4312796208530806
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.5450236966824644
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.6445497630331753
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.38906851558265765
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.3073817046565864
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.31738583597003633
            name: Cosine Map@100

BGE base Financial Matryoshka

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 dimensions
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': 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("sentence_transformers_model_id")
# Run inference
sentences = [
    'This section lists factors that decrease the seriousness of a crime.  These include age (under 18 or over 75), lack of intent, provocation by the victim, retaliation for a serious offense, confession and remorse, surrender to authorities, compensation to the victim, diminished capacity, insignificant harm, assistance in the judicial process, confession with a promise of no future crime, and crimes committed under duress.',
    'What constitutes "lack of intent" in this context?',
    'What are the exceptions to the prohibition on property transactions in marriage?',
]
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_128
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1374
cosine_accuracy@3	0.4313
cosine_accuracy@5	0.545
cosine_accuracy@10	0.6445
cosine_precision@1	0.1374
cosine_precision@3	0.1438
cosine_precision@5	0.109
cosine_precision@10	0.0645
cosine_recall@1	0.1374
cosine_recall@3	0.4313
cosine_recall@5	0.545
cosine_recall@10	0.6445
cosine_ndcg@10	0.3891
cosine_mrr@10	0.3074
cosine_map@100	0.3174

Training Details

Training Dataset

Unnamed Dataset

Size: 843 training samples
Columns: positive and anchor
Approximate statistics based on the first 843 samples:
positive anchor
type string string
details
min: 9 tokens
mean: 66.68 tokens
max: 151 tokens

min: 7 tokens
mean: 14.77 tokens
max: 39 tokens

	positive	anchor
type	string	string
details	min: 9 tokens mean: 66.68 tokens max: 151 tokens	min: 7 tokens mean: 14.77 tokens max: 39 tokens

Samples:

positive	anchor
`This law prohibits unlawful detention of individuals. It outlines penalties for unlawful confinement and obstruction of a person's movement. It also specifies a time limit for complaints related to certain offenses.`	`What is the process for reporting unlawful detention?`
`No complaint shall lie in relation to any of the offences under Section 290, after the expiry of three months from the date of commission of such offence, and in relation to any of the other offences under this Chapter, after the expiry of three months from the date of knowledge of commission of such act.`	`What are the time limits for reporting and prosecuting offenses related to animal cruelty?`
`(1) No person, being legally bound to receive a summons, process, notice, arrest warrant or order issued by the competent authority, shall abscond, with mala fide intention to avoid being served with such summons, process, notice, arrest warrant or order.`	`What is the legal definition of being "legally bound" to receive a document?`

Loss: MatryoshkaLoss with these parameters:

{
    "loss": "MultipleNegativesRankingLoss",
    "matryoshka_dims": [
        128
    ],
    "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: 1
lr_scheduler_type: cosine
warmup_ratio: 0.1
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
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: 1
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: 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_128_cosine_ndcg@10
0.5926	1	0.3891

The bold row denotes the saved checkpoint.

Framework Versions

Python: 3.10.12
Sentence Transformers: 3.3.1
Transformers: 4.47.1
PyTorch: 2.5.1+cu121
Accelerate: 0.27.0
Datasets: 3.2.0
Tokenizers: 0.21.0

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}
}