mnaguib/camembert-large-finetuned-clister

SentenceTransformer based on almanach/camembert-large

This is a sentence-transformers model finetuned from almanach/camembert-large. 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: almanach/camembert-large
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 1024 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': 512, 'do_lower_case': False}) with Transformer model: CamembertModel 
  (1): Pooling({'word_embedding_dimension': 1024, '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("sentence_transformers_model_id")
# Run inference
sentences = [
    'Le patient a été mis sous antibiothérapie adaptée (pénicilline A + aminoside).',
    'En octobre 2003, la patiente était en excellent état général.',
    "Les prélèvements de sang et d'urine sont effectués 10 heures plus tard.",
]
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

Semantic Similarity

• Dataset: sts-dev
• Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.8636
spearman_cosine	0.8677

Training Details

Training Dataset

Unnamed Dataset

• Size: 600 training samples
• Columns: sentence1, sentence2, and score
• Approximate statistics based on the first 600 samples:

  	sentence1	sentence2	score
  type	string	string	float
  details	min: 8 tokens mean: 23.23 tokens max: 85 tokens	min: 8 tokens mean: 23.08 tokens max: 90 tokens	min: 0.0 mean: 0.45 max: 1.0

• Samples:

  sentence1	sentence2	score
  L'UIV a objectivé un retard de sécrétion avec importante dilatation pyélo-calicielle et de l'uretère lombaire en amont d'un énorme calcul de l'uretère iliaque et pelvien droit (Figure 2).	L'UIV a montré une importante dilatation urétéro-pyélo-calicielle en amont d'un énorme calcul de l'uretère gauche, le coté droit était sans anomalies (Figure 6).	0.6
  1 Diminution méthadone à 80 mg TID.	7 Diminution méthadone à 20 mg TID	0.6
  L’examen clinique à l’entrée trouvait au toucher rectal une grosse vésicule séminale droite.	L'examen clinique trouvait au toucher rectal un plancher vésical fixé à gauche.	0.6

• Loss: CosineSimilarityLoss with these parameters:

  {
      "loss_fct": "torch.nn.modules.loss.MSELoss"
  }
  

Evaluation Dataset

Unnamed Dataset

• Size: 400 evaluation samples
• Columns: sentence1, sentence2, and score
• Approximate statistics based on the first 400 samples:

  	sentence1	sentence2	score
  type	string	string	float
  details	min: 8 tokens mean: 23.05 tokens max: 97 tokens	min: 7 tokens mean: 22.45 tokens max: 88 tokens	min: 0.0 mean: 0.45 max: 1.0

• Samples:

  sentence1	sentence2	score
  L’examen clinique était sans particularité.	La formule sanguine était sans particularité.	0.4
  Le bilan biologique était correct.	Le geste était complet.	0.0
  La sérologie VIH était négative.	La cytologie urinaire pyélique droite était négative.	0.2

• Loss: CosineSimilarityLoss with these parameters:

  {
      "loss_fct": "torch.nn.modules.loss.MSELoss"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_eval_batch_size: 16
• learning_rate: 1e-05
• num_train_epochs: 5
• lr_scheduler_type: cosine
• warmup_ratio: 0.1
• fp16: True
• auto_find_batch_size: True

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 8
• 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: 1e-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: 5
• max_steps: -1
• lr_scheduler_type: cosine
• 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: None
• hub_always_push: False
• 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: True
• 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
• use_liger_kernel: False
• eval_use_gather_object: False
• average_tokens_across_devices: False
• prompts: None
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	Validation Loss	sts-dev_spearman_cosine
0.1333	10	-	0.3013	0.6711
0.2667	20	-	0.2010	0.6742
0.4	30	-	0.1114	0.7249
0.5333	40	-	0.0610	0.7684
0.6667	50	-	0.0537	0.7998
0.8	60	-	0.0498	0.8245
0.9333	70	-	0.0454	0.8473
1.0667	80	-	0.0423	0.8474
1.2	90	-	0.0494	0.8625
1.3333	100	0.1041	0.0425	0.8644
1.4667	110	-	0.0391	0.8588
1.6	120	-	0.0377	0.8598
1.7333	130	-	0.0396	0.8616
1.8667	140	-	0.0378	0.8616
2.0	150	-	0.0375	0.8605
2.1333	160	-	0.0389	0.8636
2.2667	170	-	0.0387	0.8633
2.4	180	-	0.0380	0.8640
2.5333	190	-	0.0387	0.8657
2.6667	200	0.0192	0.0391	0.8655
2.8	210	-	0.0396	0.8661
2.9333	220	-	0.0373	0.8658
3.0667	230	-	0.0361	0.8662
3.2	240	-	0.0368	0.8661
3.3333	250	-	0.0370	0.8670
3.4667	260	-	0.0375	0.8681
3.6	270	-	0.0378	0.8678
3.7333	280	-	0.0374	0.8667
3.8667	290	-	0.0373	0.8662
4.0	300	0.0094	0.0373	0.8667
4.1333	310	-	0.0374	0.8676
4.2667	320	-	0.0376	0.8679
4.4	330	-	0.0376	0.8681
4.5333	340	-	0.0375	0.8680
4.6667	350	-	0.0375	0.8678
4.8	360	-	0.0374	0.8677
4.9333	370	-	0.0374	0.8677

Framework Versions

• Python: 3.12.8
• Sentence Transformers: 3.4.1
• Transformers: 4.47.1
• PyTorch: 2.5.1+cu124
• Accelerate: 1.4.0
• Datasets: 3.2.0
• 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",
}