MPAC BGE Large
This is a sentence-transformers model finetuned from BAAI/bge-large-en-v1.5 on the json dataset. It maps sentences & paragraphs to a 1024-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-large-en-v1.5
- Maximum Sequence Length: 512 tokens
- Output Dimensionality: 1024 dimensions
- Similarity Function: Cosine Similarity
- Training Dataset:
- json
- 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': 1024, '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("mp-ac/mpac-bge-large-v1.2")
# Run inference
sentences = [
'Qual a importância do NAT para o MPAC?',
'O NAT é essencial para o MPAC, fornecendo apoio técnico especializado e segurança, fortalecendo as operações de investigação e combate ao crime.',
'A Lei Complementar n.º 291 de 2014 regulamentou o NAT como um órgão auxiliar do MPAC, fortalecendo seu papel de apoio técnico e científico.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]
# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]
Evaluation
Metrics
Information Retrieval
- Datasets:
dim_768
,dim_512
,dim_256
,dim_128
anddim_64
- Evaluated with
InformationRetrievalEvaluator
Metric | dim_768 | dim_512 | dim_256 | dim_128 | dim_64 |
---|---|---|---|---|---|
cosine_accuracy@1 | 0.7778 | 0.7778 | 0.7778 | 0.7778 | 0.7778 |
cosine_accuracy@3 | 0.8889 | 0.8889 | 0.8889 | 0.8889 | 0.8889 |
cosine_accuracy@5 | 0.8889 | 0.8889 | 0.8889 | 0.8889 | 0.8889 |
cosine_accuracy@10 | 0.8889 | 1.0 | 1.0 | 1.0 | 1.0 |
cosine_precision@1 | 0.7778 | 0.7778 | 0.7778 | 0.7778 | 0.7778 |
cosine_precision@3 | 0.2963 | 0.2963 | 0.2963 | 0.2963 | 0.2963 |
cosine_precision@5 | 0.1778 | 0.1778 | 0.1778 | 0.1778 | 0.1778 |
cosine_precision@10 | 0.0889 | 0.1 | 0.1 | 0.1 | 0.1 |
cosine_recall@1 | 0.7778 | 0.7778 | 0.7778 | 0.7778 | 0.7778 |
cosine_recall@3 | 0.8889 | 0.8889 | 0.8889 | 0.8889 | 0.8889 |
cosine_recall@5 | 0.8889 | 0.8889 | 0.8889 | 0.8889 | 0.8889 |
cosine_recall@10 | 0.8889 | 1.0 | 1.0 | 1.0 | 1.0 |
cosine_ndcg@10 | 0.8333 | 0.8813 | 0.8849 | 0.8813 | 0.8849 |
cosine_mrr@10 | 0.8148 | 0.8457 | 0.8492 | 0.8457 | 0.8492 |
cosine_map@100 | 0.8249 | 0.8457 | 0.8492 | 0.8457 | 0.8492 |
Training Details
Training Dataset
json
- Dataset: json
- Size: 34 training samples
- Columns:
anchor
andpositive
- Approximate statistics based on the first 34 samples:
anchor positive type string string details - min: 8 tokens
- mean: 13.85 tokens
- max: 20 tokens
- min: 27 tokens
- mean: 53.62 tokens
- max: 76 tokens
- Samples:
anchor positive Qual é o objetivo do Simplifica?
O objetivo do Simplifica é implementar e disseminar a Linguagem Simples no Ministério Público do Estado do Acre, tornando a comunicação institucional mais acessível, clara e objetiva para todos os cidadãos.
Qual é a função do NAT no LAB-LD?
O NAT gerencia o LAB-LD, oferecendo suporte especializado em investigações financeiras para combater a lavagem de dinheiro.
O que é o NAT?
O NAT, Núcleo de Apoio Técnico, é uma unidade do Ministério Público do Estado do Acre criada em 2012 para oferecer apoio técnico, científico e de segurança aos órgãos de execução do MPAC.
- Loss:
MatryoshkaLoss
with these parameters:{ "loss": "MultipleNegativesRankingLoss", "matryoshka_dims": [ 768, 512, 256, 128, 64 ], "matryoshka_weights": [ 1, 1, 1, 1, 1 ], "n_dims_per_step": -1 }
Training Hyperparameters
Non-Default Hyperparameters
eval_strategy
: epochper_device_train_batch_size
: 32per_device_eval_batch_size
: 16gradient_accumulation_steps
: 16learning_rate
: 2e-05num_train_epochs
: 5lr_scheduler_type
: cosinewarmup_ratio
: 0.1bf16
: Truetf32
: Trueload_best_model_at_end
: Trueoptim
: adamw_torch_fusedbatch_sampler
: no_duplicates
All Hyperparameters
Click to expand
overwrite_output_dir
: Falsedo_predict
: Falseeval_strategy
: epochprediction_loss_only
: Trueper_device_train_batch_size
: 32per_device_eval_batch_size
: 16per_gpu_train_batch_size
: Noneper_gpu_eval_batch_size
: Nonegradient_accumulation_steps
: 16eval_accumulation_steps
: Nonelearning_rate
: 2e-05weight_decay
: 0.0adam_beta1
: 0.9adam_beta2
: 0.999adam_epsilon
: 1e-08max_grad_norm
: 1.0num_train_epochs
: 5max_steps
: -1lr_scheduler_type
: cosinelr_scheduler_kwargs
: {}warmup_ratio
: 0.1warmup_steps
: 0log_level
: passivelog_level_replica
: warninglog_on_each_node
: Truelogging_nan_inf_filter
: Truesave_safetensors
: Truesave_on_each_node
: Falsesave_only_model
: Falserestore_callback_states_from_checkpoint
: Falseno_cuda
: Falseuse_cpu
: Falseuse_mps_device
: Falseseed
: 42data_seed
: Nonejit_mode_eval
: Falseuse_ipex
: Falsebf16
: Truefp16
: Falsefp16_opt_level
: O1half_precision_backend
: autobf16_full_eval
: Falsefp16_full_eval
: Falsetf32
: Truelocal_rank
: 0ddp_backend
: Nonetpu_num_cores
: Nonetpu_metrics_debug
: Falsedebug
: []dataloader_drop_last
: Falsedataloader_num_workers
: 0dataloader_prefetch_factor
: Nonepast_index
: -1disable_tqdm
: Falseremove_unused_columns
: Truelabel_names
: Noneload_best_model_at_end
: Trueignore_data_skip
: Falsefsdp
: []fsdp_min_num_params
: 0fsdp_config
: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}fsdp_transformer_layer_cls_to_wrap
: Noneaccelerator_config
: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}deepspeed
: Nonelabel_smoothing_factor
: 0.0optim
: adamw_torch_fusedoptim_args
: Noneadafactor
: Falsegroup_by_length
: Falselength_column_name
: lengthddp_find_unused_parameters
: Noneddp_bucket_cap_mb
: Noneddp_broadcast_buffers
: Falsedataloader_pin_memory
: Truedataloader_persistent_workers
: Falseskip_memory_metrics
: Trueuse_legacy_prediction_loop
: Falsepush_to_hub
: Falseresume_from_checkpoint
: Nonehub_model_id
: Nonehub_strategy
: every_savehub_private_repo
: Falsehub_always_push
: Falsegradient_checkpointing
: Falsegradient_checkpointing_kwargs
: Noneinclude_inputs_for_metrics
: Falseeval_do_concat_batches
: Truefp16_backend
: autopush_to_hub_model_id
: Nonepush_to_hub_organization
: Nonemp_parameters
:auto_find_batch_size
: Falsefull_determinism
: Falsetorchdynamo
: Noneray_scope
: lastddp_timeout
: 1800torch_compile
: Falsetorch_compile_backend
: Nonetorch_compile_mode
: Nonedispatch_batches
: Nonesplit_batches
: Noneinclude_tokens_per_second
: Falseinclude_num_input_tokens_seen
: Falseneftune_noise_alpha
: Noneoptim_target_modules
: Nonebatch_eval_metrics
: Falseprompts
: Nonebatch_sampler
: no_duplicatesmulti_dataset_batch_sampler
: proportional
Training Logs
Epoch | Step | dim_768_cosine_ndcg@10 | dim_512_cosine_ndcg@10 | dim_256_cosine_ndcg@10 | dim_128_cosine_ndcg@10 | dim_64_cosine_ndcg@10 |
---|---|---|---|---|---|---|
1.0 | 1 | 0.7368 | 0.7368 | 0.7222 | 0.6686 | 0.7222 |
2.0 | 2 | 0.8128 | 0.7738 | 0.7292 | 0.7738 | 0.7702 |
3.0 | 3 | 0.8256 | 0.8258 | 0.8542 | 0.8800 | 0.8591 |
4.0 | 4 | 0.8333 | 0.8258 | 0.8704 | 0.8813 | 0.8829 |
5.0 | 5 | 0.8333 | 0.8813 | 0.8849 | 0.8813 | 0.8849 |
- The bold row denotes the saved checkpoint.
Framework Versions
- Python: 3.12.7
- Sentence Transformers: 3.3.1
- Transformers: 4.41.2
- PyTorch: 2.5.1+cu124
- Accelerate: 1.1.1
- Datasets: 3.1.0
- Tokenizers: 0.19.1
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",
}
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}
}
- Downloads last month
- 40
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Model tree for mp-ac/mpac-bge-large-v1.2
Base model
BAAI/bge-large-en-v1.5Evaluation results
- Cosine Accuracy@1 on dim 768self-reported0.778
- Cosine Accuracy@3 on dim 768self-reported0.889
- Cosine Accuracy@5 on dim 768self-reported0.889
- Cosine Accuracy@10 on dim 768self-reported0.889
- Cosine Precision@1 on dim 768self-reported0.778
- Cosine Precision@3 on dim 768self-reported0.296
- Cosine Precision@5 on dim 768self-reported0.178
- Cosine Precision@10 on dim 768self-reported0.089
- Cosine Recall@1 on dim 768self-reported0.778
- Cosine Recall@3 on dim 768self-reported0.889