pyrac/rse_gestion_durable

MPNet base trained on AllNLI triplets

This is a sentence-transformers model finetuned from pyrac/rse_engagement_des_collaborateurs. 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: pyrac/rse_engagement_des_collaborateurs
• Maximum Sequence Length: 128 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity
• 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': 128, 'do_lower_case': False}) with Transformer model: XLMRobertaModel 
  (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("pyrac/rse_gestion_durable")
# Run inference
sentences = [
    'Petit plus pour le caractère refuge LPO de l’hotel.',
    "L'établissement met en place des protocoles de sécurité au travail qui garantissent un environnement sain pour tous",
    'Parking pratique avec un bon rapport qualité-prix.',
]
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

Triplet

• Datasets: all-nli-dev and all-nli-test
• Evaluated with TripletEvaluator

Metric	all-nli-dev	all-nli-test
cosine_accuracy	1.0	1.0

Training Details

Training Dataset

Unnamed Dataset

• Size: 132,020 training samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 4 tokens mean: 22.06 tokens max: 90 tokens	min: 4 tokens mean: 19.57 tokens max: 81 tokens	min: 6 tokens mean: 18.83 tokens max: 31 tokens

• Samples:

  anchor	positive	negative
  Engagement RSE palpable, mais trop de règles vertes imposées.	Les informations sur leurs pratiques responsables sont quasi inexistantes.	Cette chambre était extrêmement décevante, elle ne correspondait absolument pas à nos besoins.
  Je suis déçu qu'aucun label environnemental comme Clef verte ne soit visible dans cet hôtel	La mise en avant de leurs pratiques éthiques est impressionnante.	Accès mal indiqué et compliqué.
  Le bien-être des employés est clairement une priorité ici avec des pratiques conformes aux dispositions légales	Ils ne sont pas aussi transparents qu'ils le prétendent.	La chambre était trop vieille et usée, ça a gâché notre séjour.

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim"
  }
  

Evaluation Dataset

Unnamed Dataset

• Size: 16,502 evaluation samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 4 tokens mean: 21.43 tokens max: 90 tokens	min: 5 tokens mean: 19.94 tokens max: 90 tokens	min: 6 tokens mean: 19.08 tokens max: 31 tokens

• Samples:

  anchor	positive	negative
  J'ai trouvé que cet hôtel avec le label Clef verte est un bel exemple d'engagement environnemental	personnels non-formés et mal payés, sous-traitance à gogo	Pas assez d'espace pour les manœuvres, surtout en heures de pointe.
  Je ne vois pas de résultats concrets de leur engagement écologique.	L'hôtel manque de transparence sur ses engagements en RSE.	On nous a placé dans une chambre qui ne correspondait vraiment pas à ce que l’on avait réservé.
  Les conditions de sécurité au travail sont irréprochables et l'environnement est sain pour les employés et les clients	RSE exemplaire, mais règles environnementales oppressives.	Vraiment déçu d’avoir eu cette chambre, ce n’était pas du tout ce qu’on s’attendait.

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 64
• per_device_eval_batch_size: 64
• num_train_epochs: 1
• warmup_ratio: 0.1
• bf16: True
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 64
• per_device_eval_batch_size: 64
• 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: 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: 1
• max_steps: -1
• lr_scheduler_type: linear
• 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: True
• 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
• 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
• 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: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	Validation Loss	all-nli-dev_cosine_accuracy	all-nli-test_cosine_accuracy
0.0485	100	4.2915	4.1299	1.0	-
0.0969	200	4.1578	4.1253	1.0	-
0.1454	300	4.1509	4.1237	1.0	-
0.1939	400	4.1465	4.1006	1.0	-
0.2424	500	4.1224	4.0881	1.0	-
0.2908	600	4.1065	4.0597	1.0	-
0.3393	700	4.0901	4.0488	1.0	-
0.3878	800	4.0862	4.0355	1.0	-
0.4363	900	4.0732	4.0352	1.0	-
0.4847	1000	4.0681	4.0271	1.0	-
0.5332	1100	4.0574	4.0270	1.0	-
0.5817	1200	4.0583	4.0235	1.0	-
0.6302	1300	4.0566	4.0180	1.0	-
0.6786	1400	4.048	4.0180	1.0	-
0.7271	1500	4.046	4.0105	1.0	-
0.7756	1600	4.0403	4.0128	1.0	-
0.8240	1700	4.0471	4.0084	1.0	-
0.8725	1800	4.0455	4.0082	1.0	-
0.9210	1900	4.0328	4.0051	1.0	-
0.9695	2000	4.0417	4.0033	1.0	-
-1	-1	-	-	-	1.0

Framework Versions

• Python: 3.12.3
• Sentence Transformers: 3.4.1
• Transformers: 4.48.3
• PyTorch: 2.6.0+cu124
• Accelerate: 1.3.0
• 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",
}

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}
}