This is a sentence-transformers model finetuned from sentence-transformers/all-MiniLM-L6-v2 on the train and test datasets. It maps sentences & paragraphs to a 384-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-MiniLM-L6-v2
Maximum Sequence Length: 256 tokens
Output Dimensionality: 384 dimensions
Similarity Function: Cosine Similarity
Training Datasets:
- train
- test

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': 256, 'do_lower_case': False, 'architecture': 'BertModel'})
  (1): Pooling({'word_embedding_dimension': 384, '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("Albertdebeauvais/all-MiniLM-L6-v2_bibliographie")
# Run inference
sentences = [
    'RITZ-GUILBERT, Anne (2018), "Les modèles du \'Bréviaire de Marie de Savoie\' par le Maître des \'Vitae Imperatorum\'", dans BORLÉE, Denise (éd.), TERRIER ALIFERIS, Laurence (éd.), Les modèles dans l\'art du Moyen Âge (XIIe-XVe siècles), Turnhout, Brepols (Répertoire iconographique de la littérature du Moyen Âge. Les études du RILMA, 10), p. 109-120',
    'GIL, Marc (2018), "Sources et circulation des modèles dans les arts figurés champenois, vers 1160-1180 : le cas de Notre-Dame-en-Vaux à Châlons-en-Champagne", dans BORLÉE, Denise (éd.), TERRIER ALIFERIS, Laurence (éd.), Les modèles dans l\'art du Moyen Âge (XIIe-XVe siècles), Turnhout, Brepols (Répertoire iconographique de la littérature du Moyen Âge. Les études du RILMA, 10), p. 179-192',
    "LEMAÎTRE, Jean-Loup (éd.) (2005), Un calendrier retrouvé : le calendrier des Heures de Saint-Pierre-du-Queyroix, Ussel, Musée du pays d'Ussel",
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.5206, 0.1709],
#         [0.5206, 1.0000, 0.2801],
#         [0.1709, 0.2801, 1.0000]])

Evaluation

Metrics

Binary Classification

Dataset: eval
Evaluated with BinaryClassificationEvaluator

Metric	Value
cosine_accuracy	0.9517
cosine_accuracy_threshold	0.8323
cosine_f1	0.9499
cosine_f1_threshold	0.8267
cosine_precision	0.9391
cosine_recall	0.961
cosine_ap	0.9647
cosine_mcc	0.9036

Training Details

Training Datasets

train

Dataset: train
Size: 26,107 training samples
Columns: text1, text2, and label
Approximate statistics based on the first 1000 samples:
text1 text2 label
type string string int
details
min: 5 tokens
mean: 66.2 tokens
max: 200 tokens
min: 4 tokens
mean: 63.27 tokens
max: 199 tokens
0: ~53.50%
1: ~46.50%

Samples:

text1	text2	label
`FORONDA, François (dir.), BARRALIS, Christine (dir.), SÈRE, Bénédicte (dir.) (2010), Violences souveraines au Moyen Âge. Travaux d'une école historique, Paris (Le nœud gordien)`	`FORONDA, François (2010), "Une image de la violence d'Etat française : la mort de Pierre Ier de Castille", dans FORONDA, François (dir.), BARRALIS, Christine (dir.), SÈRE, Bénédicte (dir.), Violences souveraines au Moyen Âge. Travaux d'une école historique, Paris (Le nœud gordien), p. 249-259`	`0`
`ORTOLEVA, Vincenzo (1994), "La cosiddetta tradizione "epitomata della Mulomedicina" di Vegezio. Recensio deterior o tradizionz indiretta ?", Revue d'histoire des textes, 24, p. 271-274`	`TOSCANO, Gennaro (1995), "Il Maestro di Isabella di Chiaromonte : note sulla miniatura a Napoli a metà Quattracento", Artes, 3, p. 34-45`	`0`
`pp. XIV, 259-262, 264-265 John LOWDEN, The Making of the Bibles moralisées. T. 2 : The Book of Ruth, University Park, The Pennsylvania State University, 2000 Mss. [ 4° Impr. 2422 (2)`	`LOWDEN, John (2000), The Making of the Bibles Moralisées : I. The manuscripts; II. The book of Ruth, University Park (PA), The Pennsylvania State University Press`	`1`

Loss: OnlineContrastiveLoss

test

Dataset: test
Size: 808 training samples
Columns: text1, text2, and label
Approximate statistics based on the first 808 samples:
text1 text2 label
type string string int
details
min: 12 tokens
mean: 67.31 tokens
max: 195 tokens
min: 11 tokens
mean: 63.32 tokens
max: 195 tokens
0: ~52.35%
1: ~47.65%

Samples:

text1	text2	label
`Pfeffer, Wendy, The Oxford Companion to Chaucer, Oxford, Oxford University Press, 2003.`	`Eglal Doss-Quinby, Joan Tasker-Grimbert, Wendy Pfeffer et Elizabeth Aubrey, Song of the Women Trouvères, New Haven/London, Yale University Press, 2001.`	`0`
`Rêves et vie spirituelle d'après Evagre le Pontique, Jérusalem, Presses Universitaires de France, 1969.`	`(1969), « Les songes et la vie quotidienne dans l'Antiquité tardive », Jérusalem, éd. Presses Universitaires de France.`	`0`
`HUCHER, Eugène (éd.) (1875-1878), Le Saint Graal, ou Le Joseph d'Arimathie, première branche des romans de la Table ronde publié d'après des textes et des documents inédits, Le Mans, Ed. Monnoyer, 3 volumes`	`1875-1878, Le Saint Graal, Ou Le Joseph D'arimathie, Première Branche Des Romans De La Table Ronde Publié D'après Des Textes Et Des Documents Inédits`	`1`

Loss: OnlineContrastiveLoss

Evaluation Dataset

Unnamed Dataset

Size: 808 evaluation samples
Columns: text1, text2, and label
Approximate statistics based on the first 808 samples:
text1 text2 label
type string string int
details
min: 12 tokens
mean: 67.31 tokens
max: 195 tokens
min: 11 tokens
mean: 63.32 tokens
max: 195 tokens
0: ~52.35%
1: ~47.65%

Samples:

text1	text2	label
`Pfeffer, Wendy, The Oxford Companion to Chaucer, Oxford, Oxford University Press, 2003.`	`Eglal Doss-Quinby, Joan Tasker-Grimbert, Wendy Pfeffer et Elizabeth Aubrey, Song of the Women Trouvères, New Haven/London, Yale University Press, 2001.`	`0`
`Rêves et vie spirituelle d'après Evagre le Pontique, Jérusalem, Presses Universitaires de France, 1969.`	`(1969), « Les songes et la vie quotidienne dans l'Antiquité tardive », Jérusalem, éd. Presses Universitaires de France.`	`0`
`HUCHER, Eugène (éd.) (1875-1878), Le Saint Graal, ou Le Joseph d'Arimathie, première branche des romans de la Table ronde publié d'après des textes et des documents inédits, Le Mans, Ed. Monnoyer, 3 volumes`	`1875-1878, Le Saint Graal, Ou Le Joseph D'arimathie, Première Branche Des Romans De La Table Ronde Publié D'après Des Textes Et Des Documents Inédits`	`1`

Loss: OnlineContrastiveLoss

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: epoch
per_device_train_batch_size: 160
per_device_eval_batch_size: 160
learning_rate: 3e-05
warmup_ratio: 0.03

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: epoch
prediction_loss_only: True
per_device_train_batch_size: 160
per_device_eval_batch_size: 160
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: 3e-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: 3
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: {}
warmup_ratio: 0.03
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: 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}
parallelism_config: 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: 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: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: proportional
router_mapping: {}
learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss	Validation Loss	eval_cosine_ap
-1	-1	-	-	0.5884
0.0353	6	7.9026	-	-
0.0706	12	7.7888	-	-
0.1059	18	7.0352	-	-
0.1412	24	6.3592	-	-
0.1765	30	5.8148	-	-
0.2118	36	4.9098	-	-
0.2471	42	5.1715	-	-
0.2824	48	4.0856	-	-
0.3176	54	4.3722	-	-
0.3529	60	4.0175	-	-
0.3882	66	3.9427	-	-
0.4235	72	3.4966	-	-
0.4588	78	3.5505	-	-
0.4941	84	3.2389	-	-
0.5294	90	3.5375	-	-
0.5647	96	3.0543	-	-
0.6	102	3.0486	-	-
0.6353	108	2.5424	-	-
0.6706	114	2.9492	-	-
0.7059	120	3.353	-	-
0.7412	126	2.7673	-	-
0.7765	132	2.9456	-	-
0.8118	138	2.3598	-	-
0.8471	144	2.5187	-	-
0.8824	150	2.2102	-	-
0.9176	156	2.675	-	-
0.9529	162	2.1735	-	-
0.9882	168	2.4117	-	-
1.0	170	-	2.0486	0.9545
1.0235	174	1.8135	-	-
1.0588	180	2.1022	-	-
1.0941	186	1.7459	-	-
1.1294	192	1.7129	-	-
1.1647	198	1.7023	-	-
1.2	204	1.8	-	-
1.2353	210	1.6906	-	-
1.2706	216	2.0856	-	-
1.3059	222	1.7216	-	-
1.3412	228	1.8287	-	-
1.3765	234	2.2071	-	-
1.4118	240	1.8617	-	-
1.4471	246	1.8148	-	-
1.4824	252	1.6976	-	-
1.5176	258	1.4774	-	-
1.5529	264	1.8896	-	-
1.5882	270	1.8389	-	-
1.6235	276	2.2744	-	-
1.6588	282	1.5614	-	-
1.6941	288	1.3118	-	-
1.7294	294	1.6211	-	-
1.7647	300	1.3294	-	-
1.8	306	2.2436	-	-
1.8353	312	1.6333	-	-
1.8706	318	1.6046	-	-
1.9059	324	1.5298	-	-
1.9412	330	1.7025	-	-
1.9765	336	1.4742	-	-
2.0	340	-	1.5898	0.9664
2.0118	342	1.5415	-	-
2.0471	348	1.1568	-	-
2.0824	354	1.3209	-	-
2.1176	360	1.2234	-	-
2.1529	366	1.7336	-	-
2.1882	372	1.382	-	-
2.2235	378	1.665	-	-
2.2588	384	1.2707	-	-
2.2941	390	1.1796	-	-
2.3294	396	1.6894	-	-
2.3647	402	1.06	-	-
2.4	408	1.0879	-	-
2.4353	414	1.2806	-	-
2.4706	420	1.6574	-	-
2.5059	426	1.5029	-	-
2.5412	432	1.3803	-	-
2.5765	438	1.2059	-	-
2.6118	444	1.7823	-	-
2.6471	450	1.2976	-	-
2.6824	456	1.6891	-	-
2.7176	462	0.9401	-	-
2.7529	468	1.1141	-	-
2.7882	474	1.1229	-	-
2.8235	480	1.137	-	-
2.8588	486	1.5186	-	-
2.8941	492	1.4301	-	-
2.9294	498	1.4644	-	-
2.9647	504	0.9985	-	-
3.0	510	0.6255	1.5778	0.9647

Framework Versions

Python: 3.9.21
Sentence Transformers: 5.1.0
Transformers: 4.56.1
PyTorch: 2.8.0+cu129
Accelerate: 1.10.1
Datasets: 4.1.0
Tokenizers: 0.22.0

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",
}

	text1	text2	label
type	string	string	int
details	min: 5 tokens mean: 66.2 tokens max: 200 tokens	min: 4 tokens mean: 63.27 tokens max: 199 tokens	0: ~53.50% 1: ~46.50%