This is a sentence-transformers model finetuned from sentence-transformers/all-mpnet-base-v2 on the json 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:
- json

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.

python

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("stephenhib/all-mpnet-base-v2-patabs-1epoc-batch32-100000")
# Run inference
sentences = [
    '<p id="pa01" num="0001">An application apparatus (100) includes: an application needle (24) that applies, to a target, an application material having its viscosity changing under shear; a drive unit (90) that moves the application needle (24) up and down; and a controller (80) that controls the drive unit (90) to move the application needle such that shear is applied to the application material at a shear speed depending on a type of the application material and depending on a target application amount or a target application diameter.<img id="iaf01" file="imgaf001.tif" wi="78" he="56" img-content="drawing" img-format="tif"/></p>',
    'COATING APPARATUS AND COATING METHOD',
    'Electric motor',
]
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: sentence-transformers/all-mpnet-base-v2
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.592
cosine_accuracy@3	0.711
cosine_accuracy@5	0.751
cosine_accuracy@10	0.814
cosine_precision@1	0.592
cosine_precision@3	0.237
cosine_precision@5	0.1502
cosine_precision@10	0.0814
cosine_recall@1	0.592
cosine_recall@3	0.711
cosine_recall@5	0.751
cosine_recall@10	0.814
cosine_ndcg@10	0.6988
cosine_mrr@10	0.6625
cosine_map@100	0.6665

Training Details

Training Dataset

json

Dataset: json
Size: 807,656 training samples
Columns: positive and anchor
Approximate statistics based on the first 1000 samples:
positive anchor
type string string
details
min: 45 tokens
mean: 237.14 tokens
max: 384 tokens
min: 3 tokens
mean: 12.34 tokens
max: 101 tokens

Samples:

positive	anchor
`[object Object]`	`IMAGE FUSION METHOD AND DEVICE`
`[object Object]`	`METHOD FOR THE DIAGNOSTIC AND/OR PROGNOSTIC ASSESSMENT OF ACUTE-ON-CHRONIC LIVER FAILURE SYNDROME IN PATIENTS WITH LIVER DISORDERS`
`[object Object]`	`Air purging pressure regulating valve`

Loss: MultipleNegativesRankingLoss with these parameters:
json
1{ 2 "scale": 20.0, 3 "similarity_fct": "cos_sim" 4}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 4
per_device_eval_batch_size: 2
learning_rate: 2e-05
num_train_epochs: 1
warmup_ratio: 0.1
bf16: True
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: 4
per_device_eval_batch_size: 2
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: 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: 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: True
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	sentence-transformers/all-mpnet-base-v2_cosine_map@100
0.032	100	0.1433	0.6217
0.064	200	0.0953	0.6447
0.096	300	0.1084	0.6612
0.128	400	0.0817	0.6546
0.16	500	0.0768	0.6512
0.192	600	0.0779	0.6466
0.224	700	0.0709	0.6594
0.256	800	0.0813	0.6441
0.288	900	0.0597	0.6454
0.32	1000	0.0744	0.6496
0.352	1100	0.0669	0.6608
0.384	1200	0.0657	0.6566
0.416	1300	0.0489	0.6660
0.448	1400	0.0643	0.6597
0.48	1500	0.0593	0.6587
0.512	1600	0.0598	0.6613
0.544	1700	0.0737	0.6570
0.576	1800	0.0661	0.6655
0.608	1900	0.0499	0.6613
0.64	2000	0.0641	0.6616
0.672	2100	0.0679	0.6662
0.704	2200	0.0521	0.6715
0.736	2300	0.0569	0.6651
0.768	2400	0.0507	0.6679
0.8	2500	0.0405	0.6678
0.832	2600	0.0548	0.6690
0.864	2700	0.0403	0.6692
0.896	2800	0.0613	0.6649
0.928	2900	0.0485	0.6673
0.96	3000	0.0495	0.6674
0.992	3100	0.0546	0.6665

Framework Versions

Python: 3.11.9
Sentence Transformers: 3.2.1
Transformers: 4.45.2
PyTorch: 2.3.1.post300
Accelerate: 1.0.1
Datasets: 3.0.1
Tokenizers: 0.20.1

Citation

BibTeX

Sentence Transformers

bibtex

@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

bibtex

@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}
}

	positive	anchor
type	string	string
details	min: 45 tokens mean: 237.14 tokens max: 384 tokens	min: 3 tokens mean: 12.34 tokens max: 101 tokens