julian-schelb/SPhilBerta-latin-intertextuality

SentenceTransformer based on bowphs/SPhilBerta

This is a sentence-transformers model finetuned from bowphs/SPhilBerta. 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: bowphs/SPhilBerta
• Maximum Sequence Length: 128 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': 128, '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("julian-schelb/SPhilBerta-latin-intertextuality")
# Run inference
sentences = [
    'Query: cuius rei argumenta sunt nummi ueteres, in quibus est cum duplici fronte Ianus et in altera parte nauis, sicut idem poeta subiecit: at bona posteritas puppem formauit in aere hospitis aduentum testificata dei.',
    'Candidate: at bona posteritas puppem formauit in aere,  hospitis aduentum testificata dei.',
    'Candidate: cum enim paupertatis una eademque sit vis, quidnam dici potest, quam ob rem  Fabricio tolerabilis ea fuerit, alii negent se ferre posse?',
]
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: 1,122 training samples
• Columns: query, match, and label
• Approximate statistics based on the first 1000 samples:

  	query	match	label
  type	string	string	int
  details	min: 10 tokens mean: 41.73 tokens max: 125 tokens	min: 6 tokens mean: 31.92 tokens max: 128 tokens	0: ~90.80% 1: ~9.20%

• Samples:

  query	match	label
  Query: nam si, ut ipsa petit maiestas cognita rerum, - dicendum est, deus ille fuit, deus, inclute Memmi	Candidate: ipse deum genitor caelo terrisque canebat.	0
  Query: uerum id genus sacrificii ab Hercule, cum ex Hispania rediret, dicitur esse sublatum, ritu tamen permanente ut pro ueris hominibus imagines iacerentur e scirpo, ut Ouidius in Fastis docet: donec in haec uenit Tirynthius arua, quotannis tristia Leucadio sacra peracta modo.	Candidate: Vesper adest, iuuenes, consurgite: Vesper Olympo expectata diu uix tandem lumina tollit.	0
  Query: Quod in principio templi Ezechielis debui dicere, nunc praepostero ordine in fine dicturus sum, illius uersiculi memor: Hic labor ille domus, et inextricabilis error.	Candidate: Tu, ut antea fecisti, velim, si qui erunt ad quos aliquid scribendum a me existimes, ipse conficias.	0

• Loss: OnlineContrastiveLoss

Evaluation Dataset

Unnamed Dataset

• Size: 330 evaluation samples
• Columns: query, match, and label
• Approximate statistics based on the first 330 samples:

  	query	match	label
  type	string	string	int
  details	min: 18 tokens mean: 38.7 tokens max: 74 tokens	min: 7 tokens mean: 31.88 tokens max: 105 tokens	0: ~90.91% 1: ~9.09%

• Samples:

  query	match	label
  Query: quanti enim operantur terram et exercent uomere et tamen multis inpedientibus causis egestate conficiuntur et penuria!	Candidate: hoc cum ceterae gentes sic arbitrantur, tum ipsis Siculis ita persuasum est ut in animis eorum insitum atque innatum esse videatur.	0
  Query: nam quia tum Corybautes galearum tinnitibus et scutorum pulsibus uagitum pueri texerant , nunc imago rei refertur in sacris, sed pro galeis cymbala, pro scutis tympana feriuntur, ne puerum uagientem Saturnus exaudiat.	Candidate: quod si essent falsae notitiae enim notitias appellare tu videbare - si igitur essent eae falsae aut eius modi visis inpressae qualia visa a falsis discerni non possent, quo tandem his modo uteremur, quo modo autem quid cuique rei consentaneum esset quid repugnaret videremus?	0
  Query: De quibus Uirgilius: Omnigenumque deum monstra, et latrator Anubis.	Candidate: haec igitur minui, cum sint detrita, videmus.	0

• Loss: OnlineContrastiveLoss

Training Hyperparameters

Non-Default Hyperparameters

• overwrite_output_dir: True
• eval_strategy: steps
• per_device_train_batch_size: 32
• learning_rate: 2e-05
• weight_decay: 0.01
• num_train_epochs: 4
• warmup_steps: 448
• prompts: {'query': 'Query: ', 'match': 'Candidate: '}

All Hyperparameters

Click to expand

• overwrite_output_dir: True
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 8
• 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: 0.01
• 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: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.0
• warmup_steps: 448
• 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: 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: None
• hub_always_push: False
• hub_revision: None
• 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
• 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
• liger_kernel_config: None
• eval_use_gather_object: False
• average_tokens_across_devices: False
• prompts: {'query': 'Query: ', 'match': 'Candidate: '}
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: proportional

Framework Versions

• Python: 3.10.8
• Sentence Transformers: 4.1.0
• Transformers: 4.53.0
• PyTorch: 2.7.1+cu126
• Accelerate: 1.4.0
• Datasets: 3.3.2
• Tokenizers: 0.21.0

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",
}