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

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("carnival13/all-mpnet-base-v2-modulepred")
# Run inference
sentences = [
    'module: bread ambient group: bread ambient supergroup: food ambient example descriptions: 1 war 3 toastie 400 g cc 90 varburtons bread tovis snelwrspmpkin 800 g warbutons medium bread spk giant crumpets z hovis med wht 600 g sandwich thins 5 pk warb pk crumpets mission plain tortilla 25 cm warburtons 4 protein thin bagels hovis soft wet med hovis wholemefl pataks pappadums 6 pk warb so bth disc pappajuns',
    'retailer: crispcorner description: kingsmill 5050 medius bread 800 g',
    'retailer: vitalveg description: ready to eat prun',
]
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	0.4988
cosine_accuracy@3	0.6342
cosine_accuracy@5	0.7102
cosine_accuracy@10	0.7838
cosine_precision@1	0.4988
cosine_precision@3	0.2114
cosine_precision@5	0.142
cosine_precision@10	0.0784
cosine_recall@1	0.4988
cosine_recall@3	0.6342
cosine_recall@5	0.7102
cosine_recall@10	0.7838
cosine_ndcg@10	0.6324
cosine_mrr@10	0.585
cosine_map@100	0.591

Training Details

Training Dataset

csv

Dataset: csv
Size: 505,654 training samples
Columns: query and full_doc
Approximate statistics based on the first 1000 samples:
query full_doc
type string string
details
min: 10 tokens
mean: 14.8 tokens
max: 23 tokens
min: 83 tokens
mean: 115.71 tokens
max: 176 tokens

Samples:

query	full_doc
`retailer: vitalveg description: twin xira`	`module: chocolate single variety group: chocolate chocolate substitutes supergroup: biscuits & confectionery & snacks example descriptions: milky way twin 43 crml prtzlarum rai galaxy mnstr pipnut 34 g dark pb cup nest mnch foge p nestle smarties shar dark choc chun x 10 pk kinder bueno 1 dr oetker 72 da poppets choc offee pouch yorkie biscuit zpk haltesers truffles bog cadbury mini snowballs p terrys choc orange 3435 g galaxy fusion dark 704 100 g`
`retailer: freshnosh description: mab pop sockt`	`module: clothing & personal accessories group: clothing & personal accessories supergroup: clothing & personal accessories example descriptions: pk blue trad ging 40 d 3 pk opaque tight t 74 green cali jogger ss animal swing yb denim stripe pump aw 21 ff vest aw 21 girls 5 pk lounge toplo sku 1 pk fleecy tight knitted pom hat pk briefs timeless double pom pomkids hat cute face twosie sku coral jersey str pun faded petrol t 32 seamfree waist c`
`retailer: nourify description: bts prwn ckt swch`	`module: bread sandwiches filled rolls wraps group: bread fresh fixed weight supergroup: food perishable example descriptions: us chicken may hamche sw jo dbs allbtr pp st 4 js baconfree ran posh cheesy bea naturify cb swich sp eggcress f cpdfeggbacon js cheeseonion sv duck wrap reduced price takeout egg mayo sandwich 7 takeout cheeseonion s wich 2 ad leicester plough bts cheese pman 2 1 cp bacon chese s`

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: 16
learning_rate: 2e-05
num_train_epochs: 1
warmup_ratio: 0.1
fp16: 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: 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: 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: 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
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
0.0016	100	1.6195	0.2567
0.0032	200	1.47	0.3166
0.0047	300	1.2703	0.3814
0.0063	400	1.1335	0.4495
0.0079	500	0.9942	0.4827
0.0095	600	0.9004	0.5058
0.0111	700	0.8838	0.5069
0.0016	100	0.951	0.5197
0.0032	200	0.9597	0.5323
0.0047	300	0.9241	0.5406
0.0063	400	0.8225	0.5484
0.0079	500	0.7961	0.5568
0.0095	600	0.7536	0.5621
0.0111	700	0.7387	0.5623
0.0127	800	0.7716	0.5746
0.0142	900	0.7921	0.5651
0.0158	1000	0.7744	0.5707
0.0174	1100	0.8021	0.5770
0.0190	1200	0.732	0.5756
0.0206	1300	0.764	0.5798
0.0221	1400	0.7726	0.5873
0.0237	1500	0.6676	0.5921
0.0253	1600	0.6851	0.5841
0.0269	1700	0.7404	0.5964
0.0285	1800	0.6798	0.5928
0.0301	1900	0.6485	0.5753
0.0316	2000	0.649	0.5839
0.0332	2100	0.6739	0.5891
0.0348	2200	0.6616	0.6045
0.0364	2300	0.6287	0.5863
0.0380	2400	0.6602	0.5898
0.0396	2500	0.5667	0.5910

Framework Versions

Python: 3.10.14
Sentence Transformers: 3.1.1
Transformers: 4.44.2
PyTorch: 2.4.0+cu124
Accelerate: 0.33.0
Datasets: 2.21.0
Tokenizers: 0.19.1

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}
}

	query	full_doc
type	string	string
details	min: 10 tokens mean: 14.8 tokens max: 23 tokens	min: 83 tokens mean: 115.71 tokens max: 176 tokens