README.md

metadata

base_model: PlanTL-GOB-ES/roberta-base-bne
datasets: []
language: []
library_name: sentence-transformers
metrics:
  - cosine_accuracy@1
  - cosine_accuracy@3
  - cosine_accuracy@5
  - cosine_accuracy@10
  - cosine_precision@1
  - cosine_precision@3
  - cosine_precision@5
  - cosine_precision@10
  - cosine_recall@1
  - cosine_recall@3
  - cosine_recall@5
  - cosine_recall@10
  - cosine_ndcg@10
  - cosine_mrr@10
  - cosine_map@100
pipeline_tag: sentence-similarity
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:4173
  - loss:MatryoshkaLoss
  - loss:MultipleNegativesRankingLoss
widget:
  - source_sentence: >-
      L'Ajuntament de Sitges ofereix a aquelles famílies que acompleixin els
      requisits establerts, ajuts per al pagament de la quota del servei i de la
      quota del menjador dels infants matriculats a les Llars d'Infants
      Municipals ( 0-3 anys).
    sentences:
      - >-
        Quin és l'objectiu principal de les subvencions per a projectes i
        activitats de l'àmbit turístic?
      - >-
        Quin és el procediment per a obtenir una llicència per a disposar d'una
        parada en un mercat setmanal?
      - >-
        Quin és el paper de l'Ajuntament de Sitges en la quota del menjador de
        les Llars d'Infants Municipals?
  - source_sentence: >-
      Es tracta de la sol·licitud de permís municipal per poder utilitzar de
      forma privativa una zona de la via pública per instal·lacions d’atraccions
      i venda en fires, amb independència de les possibles afectacions a la via
      pública...
    sentences:
      - Quin és el tipus de permís que es sol·licita?
      - Quin és el paper de l'Ajuntament en aquest tràmit?
      - >-
        Quin és el resultat de la llicència per a la constitució d'un règim de
        propietat horitzontal en relació amb l’escriptura de divisió
        horitzontal?
  - source_sentence: >-
      Totes les persones que resideixen a Espanya estan obligades a inscriure's
      en el padró del municipi en el qual resideixen habitualment.
    sentences:
      - >-
        Quin és el benefici de l'ajut extraordinari per a la família de
        l'empleat?
      - >-
        Què passa si no es presenta la sol·licitud d'acceptació en el termini
        establert?
      - Qui està obligat a inscriure's en el Padró Municipal d'Habitants?
  - source_sentence: >-
      Les persones i entitats beneficiaries hauran de justificar la realització
      del projecte/activitat subvencionada com a màxim el dia 31 de març de
      2023.
    sentences:
      - >-
        Quin és el termini per presentar la justificació de la realització del
        projecte/activitat subvencionada?
      - Quin és el període durant el qual es poden sol·licitar els ajuts?
      - >-
        Quin és el registre on s'inscriuen les entitats d’interès ciutadà de
        Sitges?
  - source_sentence: >-
      Els establiments locals tenen un paper clau en el projecte de la targeta
      de fidelització, ja que són els que ofereixen descomptes i ofertes
      especials als consumidors que utilitzen la targeta.
    sentences:
      - >-
        Quin és el paper dels establiments locals en el projecte de la targeta
        de fidelització?
      - Quin és el paper de la via pública en aquest tràmit?
      - >-
        Quins són els tractaments que beneficien la salut de l'empleat municipal
        que s'inclouen en l'ajuda?
model-index:
  - name: SentenceTransformer based on PlanTL-GOB-ES/roberta-base-bne
    results:
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 768
          type: dim_768
        metrics:
          - type: cosine_accuracy@1
            value: 0.05172413793103448
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.125
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.1875
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.3857758620689655
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.05172413793103448
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.041666666666666664
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.037500000000000006
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.038577586206896546
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.05172413793103448
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.125
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.1875
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.3857758620689655
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.18210165785971896
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.12195368089764656
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.14624173028144724
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 512
          type: dim_512
        metrics:
          - type: cosine_accuracy@1
            value: 0.04525862068965517
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.11422413793103449
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.1810344827586207
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.38146551724137934
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.04525862068965517
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.038074712643678156
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.03620689655172414
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.03814655172413794
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.04525862068965517
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.11422413793103449
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.1810344827586207
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.38146551724137934
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.17527880528414544
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.1144422208538589
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.1390144478189839
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 256
          type: dim_256
        metrics:
          - type: cosine_accuracy@1
            value: 0.04741379310344827
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.12284482758620689
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.20043103448275862
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.39870689655172414
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.04741379310344827
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.040948275862068964
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.04008620689655173
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.03987068965517242
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.04741379310344827
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.12284482758620689
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.20043103448275862
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.39870689655172414
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.18506006244346174
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.12174072933771223
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.1456718436547049
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 128
          type: dim_128
        metrics:
          - type: cosine_accuracy@1
            value: 0.04525862068965517
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.11637931034482758
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.16810344827586207
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.38146551724137934
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.04525862068965517
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.03879310344827586
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.03362068965517242
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.03814655172413794
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.04525862068965517
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.11637931034482758
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.16810344827586207
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.38146551724137934
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.17534824322616613
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.11455682129173515
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.139297198421225
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 64
          type: dim_64
        metrics:
          - type: cosine_accuracy@1
            value: 0.03879310344827586
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.09698275862068965
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.15301724137931033
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.34698275862068967
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.03879310344827586
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.03232758620689655
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.030603448275862068
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.034698275862068965
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.03879310344827586
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.09698275862068965
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.15301724137931033
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.34698275862068967
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.1568993526090112
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.10108699370552827
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.1268371950835345
            name: Cosine Map@100

SentenceTransformer based on PlanTL-GOB-ES/roberta-base-bne

This is a sentence-transformers model finetuned from PlanTL-GOB-ES/roberta-base-bne. 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: PlanTL-GOB-ES/roberta-base-bne
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 tokens
• 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: RobertaModel 
  (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("adriansanz/sitges10242608-4ep-rerankv3-sp")
# Run inference
sentences = [
    'Els establiments locals tenen un paper clau en el projecte de la targeta de fidelització, ja que són els que ofereixen descomptes i ofertes especials als consumidors que utilitzen la targeta.',
    'Quin és el paper dels establiments locals en el projecte de la targeta de fidelització?',
    "Quins són els tractaments que beneficien la salut de l'empleat municipal que s'inclouen en l'ajuda?",
]
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: dim_768
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.0517
cosine_accuracy@3	0.125
cosine_accuracy@5	0.1875
cosine_accuracy@10	0.3858
cosine_precision@1	0.0517
cosine_precision@3	0.0417
cosine_precision@5	0.0375
cosine_precision@10	0.0386
cosine_recall@1	0.0517
cosine_recall@3	0.125
cosine_recall@5	0.1875
cosine_recall@10	0.3858
cosine_ndcg@10	0.1821
cosine_mrr@10	0.122
cosine_map@100	0.1462

Information Retrieval

• Dataset: dim_512
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.0453
cosine_accuracy@3	0.1142
cosine_accuracy@5	0.181
cosine_accuracy@10	0.3815
cosine_precision@1	0.0453
cosine_precision@3	0.0381
cosine_precision@5	0.0362
cosine_precision@10	0.0381
cosine_recall@1	0.0453
cosine_recall@3	0.1142
cosine_recall@5	0.181
cosine_recall@10	0.3815
cosine_ndcg@10	0.1753
cosine_mrr@10	0.1144
cosine_map@100	0.139

Information Retrieval

• Dataset: dim_256
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.0474
cosine_accuracy@3	0.1228
cosine_accuracy@5	0.2004
cosine_accuracy@10	0.3987
cosine_precision@1	0.0474
cosine_precision@3	0.0409
cosine_precision@5	0.0401
cosine_precision@10	0.0399
cosine_recall@1	0.0474
cosine_recall@3	0.1228
cosine_recall@5	0.2004
cosine_recall@10	0.3987
cosine_ndcg@10	0.1851
cosine_mrr@10	0.1217
cosine_map@100	0.1457

Information Retrieval

• Dataset: dim_128
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.0453
cosine_accuracy@3	0.1164
cosine_accuracy@5	0.1681
cosine_accuracy@10	0.3815
cosine_precision@1	0.0453
cosine_precision@3	0.0388
cosine_precision@5	0.0336
cosine_precision@10	0.0381
cosine_recall@1	0.0453
cosine_recall@3	0.1164
cosine_recall@5	0.1681
cosine_recall@10	0.3815
cosine_ndcg@10	0.1753
cosine_mrr@10	0.1146
cosine_map@100	0.1393

Information Retrieval

• Dataset: dim_64
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.0388
cosine_accuracy@3	0.097
cosine_accuracy@5	0.153
cosine_accuracy@10	0.347
cosine_precision@1	0.0388
cosine_precision@3	0.0323
cosine_precision@5	0.0306
cosine_precision@10	0.0347
cosine_recall@1	0.0388
cosine_recall@3	0.097
cosine_recall@5	0.153
cosine_recall@10	0.347
cosine_ndcg@10	0.1569
cosine_mrr@10	0.1011
cosine_map@100	0.1268

Training Details

Training Dataset

Unnamed Dataset

• Size: 4,173 training samples
• Columns: positive and anchor
• Approximate statistics based on the first 1000 samples:

  	positive	anchor
  type	string	string
  details	min: 10 tokens mean: 60.84 tokens max: 206 tokens	min: 10 tokens mean: 25.34 tokens max: 53 tokens

• Samples:

  positive	anchor
  L'objectiu principal de la persona coordinadora de colònia felina és garantir el benestar dels animals de la colònia.	Quin és l'objectiu principal de la persona coordinadora de colònia felina?
  Es tracta d'una sala amb capacitat per a 125 persones, equipada amb un petit escenari, sistema de sonorització, pantalla per a projeccions, camerins i serveis higiènics (WC).	Quin és el nombre de persones que pot acollir la sala d'actes del Casal Municipal de la Gent Gran de Sitges?
  Aquest ajut pretén fomentar l’associacionisme empresarial local, per tal de disposar d’agrupacions, gremis o associacions representatives de l’activitat empresarial del municipi.	Quin és el paper de les empreses en aquest ajut?

• 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: epoch
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• gradient_accumulation_steps: 16
• num_train_epochs: 4
• lr_scheduler_type: cosine
• warmup_ratio: 0.2
• bf16: True
• tf32: False
• load_best_model_at_end: True
• optim: adamw_torch_fused
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: epoch
• 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: 16
• eval_accumulation_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: 4
• max_steps: -1
• lr_scheduler_type: cosine
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.2
• 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: False
• 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: True
• 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_fused
• 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
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	dim_128_cosine_map@100	dim_256_cosine_map@100	dim_512_cosine_map@100	dim_64_cosine_map@100	dim_768_cosine_map@100
0.6130	10	10.8464	-	-	-	-	-
0.9808	16	-	0.1060	0.1088	0.1067	0.0984	0.1074
1.2261	20	3.5261	-	-	-	-	-
1.8391	30	1.4363	-	-	-	-	-
1.9617	32	-	0.1406	0.1468	0.1356	0.1395	0.1373
2.4521	40	0.5627	-	-	-	-	-
2.9425	48	-	0.1377	0.1418	0.1427	0.1322	0.1437
3.0651	50	0.2727	-	-	-	-	-
3.6782	60	0.1297	-	-	-	-	-
3.9234	64	-	0.1393	0.1457	0.139	0.1268	0.1462

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.0.1
• Transformers: 4.42.4
• PyTorch: 2.4.0+cu121
• Accelerate: 0.34.0.dev0
• Datasets: 2.21.0
• 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}
}