jebish7/mpnet-base-all-obliqa

SentenceTransformer based on sentence-transformers/all-mpnet-base-v2

This is a sentence-transformers model finetuned from sentence-transformers/all-mpnet-base-v2 on the obliqa_embed_rand 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: sentence-transformers/all-mpnet-base-v2
• Maximum Sequence Length: 384 tokens
• Output Dimensionality: 768 tokens
• Similarity Function: Cosine Similarity
• Training Dataset:
  • obliqa_embed_rand

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': 384, 'do_lower_case': False}) with Transformer model: MPNetModel 
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, '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("jebish7/mpnet-base-all-obliqa")
# Run inference
sentences = [
    'Can the ADGM clarify the specific criteria that must be met for an Authorised Person to receive approval to use the Standardised Approach or Alternative Standardised Approach for calculating the Operational Risk Capital Requirement?',
    'DocumentID: 13 | PassageID: APP7.Guidance.1. | Passage: Section 6.11 of these Rules provides that an Authorised Person in Categories 1, 2, 3A and 5 must use the Basic Indicator Approach to calculate its Operational Risk Capital Requirement, unless the firm has approval from the Regulator to use the Standardised Approach or Alternative Standardised Approach. In this App7:\na.\tthe Basic Indicator Approach is prescribed in Section A7.1;\nb.\tthe Standardised Approach is prescribed in Section A7.2; and\nc.\tthe Alternative Standardised Approach is prescribed in Section A7.3.',
    "DocumentID: 7 | PassageID: 8.8.12.(2) | Passage: The Authorised Person's annual report on its Controllers must include:\n(a)\tthe name of each Controller; and\n(b)\tthe current holding of each Controller, expressed as a percentage.",
]
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]

Training Details

Training Dataset

obliqa_embed_rand

• Dataset: obliqa_embed_rand at 8fe381e
• Size: 66,883 training samples
• Columns: question, details, and label
• Approximate statistics based on the first 1000 samples:

  	question	details	label
  type	string	string	int
  details	min: 16 tokens mean: 34.84 tokens max: 64 tokens	min: 16 tokens mean: 96.62 tokens max: 384 tokens	0: ~68.20% 1: ~31.80%

• Samples:

  question	details	label
  Does the FSRA provide any workshops, training sessions, or additional support to ensure that our Competent Persons are fully equipped to comply with the PRMS as incorporated in Chapter 12 of MKT?	DocumentID: 31	PassageID: 15)
  Are there specific ADGM guidelines on how to effectively monitor transactions to detect those that are not consistent with a Relevant Person’s knowledge of the customer?	DocumentID: 13	PassageID: APP6.A6.8.14
  In assessing a Person's investment portfolio for Professional Client status, what portfolio composition and size are generally considered to demonstrate sufficient sophistication and risk understanding?	DocumentID: 3	PassageID: 2.6.2.(a)

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Evaluation Dataset

obliqa_embed_rand

• Dataset: obliqa_embed_rand at 8fe381e

• Size: 66,883 evaluation samples

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "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: 1
• warmup_ratio: 0.1
• batch_sampler: no_duplicates

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
• torch_empty_cache_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: 1
• 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: 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: 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
• use_liger_kernel: False
• eval_use_gather_object: False
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	loss
0.0516	200	2.1914	1.6942
0.1032	400	1.4982	1.3726
0.1548	600	1.2475	1.1403
0.2065	800	1.099	1.0289
0.2581	1000	1.0372	0.9229
0.3097	1200	0.9304	0.9148
0.3613	1400	0.899	0.8668
0.4129	1600	0.8795	0.8177
0.4645	1800	0.8066	0.7916
0.5161	2000	0.776	0.7550
0.5677	2200	0.7861	0.7524
0.6194	2400	0.7482	0.7417
0.6710	2600	0.7192	0.6892
0.7226	2800	0.7115	0.6662
0.7742	3000	0.673	0.6384
0.8258	3200	0.6717	0.6307
0.8774	3400	0.6075	0.6188
0.9290	3600	0.6358	0.6011
0.9806	3800	0.5916	0.5948

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.1.1
• Transformers: 4.45.2
• PyTorch: 2.5.1+cu121
• Accelerate: 1.1.1
• Datasets: 3.1.0
• Tokenizers: 0.20.3

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",
}

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}
}

ObliQA dataset

@misc{gokhan2024regnlpactionfacilitatingcompliance,
      title={RegNLP in Action: Facilitating Compliance Through Automated Information Retrieval and Answer Generation}, 
      author={Tuba Gokhan and Kexin Wang and Iryna Gurevych and Ted Briscoe},
      year={2024},
      eprint={2409.05677},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2409.05677}, 
}