NDA_finetuned_V1 / README.md

lucagafner

Update README.md

e9a15c5 verified 5 months ago

preview code

raw

history blame contribute delete

14.6 kB

metadata

tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:45
  - loss:DenoisingAutoEncoderLoss
base_model: nlpaueb/legal-bert-base-uncased
widget:
  - source_sentence: >-
      The not apply to the prove to part of public at time it without violation
      any non-disclosure in; or already Receiving before disclosure Party
      evidenced its written records revealed Receiving by a third without
      non-disclosure favour of the Party received the Party
    sentences:
      - >-
        The receiving party will segregate Confidential Information from the
        confidential materials of third parties to prevent commingling.
      - |-
        NON-DISCLOSURE AGREEMENT (NDA)

        1.
      - >-
        The non-disclosure undertaking under this Agreement shall not apply to
        information which the Receiving Party can prove to

        have been part of public knowledge at the time the Receiving Party
        received it or became public knowledge thereafter without violation of
        any non-disclosure undertaking in favour of the Disclosing Party; or

        have been already known to the Receiving Party before disclosure by the
        Disclosing Party as evidenced by its written records or has been
        revealed to the Receiving Party by a third party without violation of a
        non-disclosure undertaking in favour of [Name disclosing party]; or

        have been developed by the Receiving Party independently of the
        information received by the Disclosing Party.
  - source_sentence: Disclosing receives Confidential Information Receiving for,,, 6,, Party
    sentences:
      - >-
        THE PARTIES AGREE AS FOLLOWS:


        The non-disclosure undertaking of the Receiving Party covers all
        information provided by the Disclosing Party, or any third party on
        behalf of the Disclosing Party, to the Receiving Party (the Confidential
        Information), whether transferred on paper, verbally, electronically, or
        by any other means or on any other media.
      - >-
        In case the Disclosing Party receives any Confidential Information from
        the Receiving Party for the purposes described in Section 2, the
        sections 3, 4, 5, 6, 7, 8 and 9 shall apply also to the Disclosing
        Party.
      - >-
        The term “person” as used in this Agreement shall be interpreted to
        include, without limitation, any other corporation, company, group,
        partnership or individual.
  - source_sentence: >-
      return to disclosing party all forms such Information in control,
      including to drawings, documents, models or any material copies thereof,
      delete would be any but hardware
    sentences:
      - >-

        PREAMBLE

        The Disclosing Party will disclose to the Receiving Party certain
        information which is non-public at the time of disclosure and considered
        confidential by the Disclosing Party.
      - >-
        Upon request, the receiving party will return or deliver to the
        disclosing party all tangible forms of such Confidential Information in
        its possession or control, including but not limited to drawings,
        specifications, documents, records, devices, models or any other
        material and copies or reproductions thereof, respectively delete
        Confidential Information that would be contained on any intangible
        format such as but not exclusively hardware.
      - >-
        Governing Law and Jurisdiction

        This Agreement is governed by and shall be construed in accordance with
        the laws of Switzerland.
  - source_sentence: “ Receiving the below “ Receiving
    sentences:
      - >-
        The “Receiving Party” is specified by the information below (see
        “Receiving Party”).
      - The Parties submit to the exclusive jurisdiction of Zug.
      - >-
        The courts at the domicile of [Name of disclosing party] shall have
        exclusive jurisdiction.
  - source_sentence: >-
      that term Agreement it the solely the purposes Section 2 Confidential
      Information at its costs will appropriate confidentiality
    sentences:
      - >-
        The “Disclosing Party” in this Agreement refers to [Name Disclosing
        Party].
      - >-
        The Parties hereto undertake to replace such invalid or ineffective
        provision by an effective/valid provision.
      - >-
        The Receiving Party hereby agrees that for the term of this Agreement it
        shall use the Confidential Information solely for the purposes described
        in Section 2 and shall keep the Confidential Information strictly
        confidential, at all times and at its own costs and will take
        appropriate steps to protect the confidentiality thereof.
pipeline_tag: sentence-similarity
library_name: sentence-transformers

SentenceTransformer based on nlpaueb/legal-bert-base-uncased

This is a sentence-transformers model finetuned from nlpaueb/legal-bert-base-uncased. 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: nlpaueb/legal-bert-base-uncased
Maximum Sequence Length: 512 tokens
Output Dimensionality: 768 dimensions
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': False}) with Transformer model: BertModel 
  (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})
)

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("lucagafner/NDA_finetuned_V1")
# Run inference

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

Unnamed Dataset

Size: 45 training samples
Columns: damaged_sentence and original_sentence
Approximate statistics based on the first 45 samples:
damaged_sentence original_sentence
type string string
details
min: 3 tokens
mean: 18.18 tokens
max: 81 tokens

min: 4 tokens
mean: 41.53 tokens
max: 183 tokens
Loss: DenoisingAutoEncoderLoss

	damaged_sentence	original_sentence
type	string	string
details	min: 3 tokens mean: 18.18 tokens max: 81 tokens	min: 4 tokens mean: 41.53 tokens max: 183 tokens

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 16
per_device_eval_batch_size: 16
num_train_epochs: 5
fp16: True

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: no
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: 5
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: {}
warmup_ratio: 0.0
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}
tp_size: 0
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: 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: batch_sampler
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss
0.3333	1	3.3557
0.6667	2	4.3068
1.0	3	3.916
1.3333	4	3.1851
1.6667	5	3.2759
2.0	6	3.1665
2.3333	7	3.1407
2.6667	8	2.6952
3.0	9	2.4053
3.3333	10	2.5579
3.6667	11	2.0525
4.0	12	2.2234
4.3333	13	1.8476
4.6667	14	2.0873
5.0	15	1.8472

Framework Versions

Python: 3.12.9
Sentence Transformers: 4.0.1
Transformers: 4.50.2
PyTorch: 2.6.0+cu124
Accelerate: 1.5.2
Datasets: 3.5.0
Tokenizers: 0.21.1

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",
}

DenoisingAutoEncoderLoss

@inproceedings{wang-2021-TSDAE,
    title = "TSDAE: Using Transformer-based Sequential Denoising Auto-Encoderfor Unsupervised Sentence Embedding Learning",
    author = "Wang, Kexin and Reimers, Nils and Gurevych, Iryna",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2021",
    month = nov,
    year = "2021",
    address = "Punta Cana, Dominican Republic",
    publisher = "Association for Computational Linguistics",
    pages = "671--688",
    url = "https://arxiv.org/abs/2104.06979",
}