software-si/kitchen-ita-nli-deberta

CrossEncoder based on cross-encoder/nli-deberta-v3-base

This is a Cross Encoder model finetuned from cross-encoder/nli-deberta-v3-base on the json dataset using the sentence-transformers library. It computes scores for pairs of texts, which can be used for text pair classification.

Model Details

Model Description

• Model Type: Cross Encoder
• Base model: cross-encoder/nli-deberta-v3-base
• Maximum Sequence Length: 512 tokens
• Number of Output Labels: 3 labels
• Training Dataset:
  • json

Model Sources

• Documentation: Sentence Transformers Documentation
• Documentation: Cross Encoder Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Cross Encoders on Hugging Face

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 CrossEncoder

# Download from the 🤗 Hub
model = CrossEncoder("software-si/kitchen-ita-nli-deberta")
# Get scores for pairs of texts
pairs = [
    ['piano cottura due zone operative, con forno a gas,', 'la cucina ha un forno integrato'],
    ['unità di cottura a induzione, sistema con forno a gas, ampiezza di novanta centimetri, con quattro piastre cottura,', 'la cucina è profonda 90 cm'],
    ['modulo cucina funzionamento a induzione, dimensione teglie di gn1/1 due moduli di cottura, dotata di forno a gas,', 'la teglie del forno hanno dimensione gn1/1'],
    ['unità di cottura modulo con forno a gas, con teglie di gn1/1 piano a induzione,', 'la cucina ha un vano aperto'],
    ['unità di cottura misura 90 centimetri di profondità, modulo disposto su vano chiuso, operativa a gas, sei bruciatori separati,', 'la cucina ha un forno integrato'],
]
scores = model.predict(pairs)
print(scores.shape)
# (5, 3)

Training Details

Training Dataset

json

• Dataset: json
• Size: 463,254 training samples
• Columns: premises, hypothesis, and labels
• Approximate statistics based on the first 1000 samples:

  	premises	hypothesis	labels
  type	string	string	int
  details	min: 37 characters mean: 103.4 characters max: 155 characters	min: 12 characters mean: 33.04 characters max: 50 characters	0: ~48.70% 1: ~47.10% 2: ~4.20%

• Samples:

  premises	hypothesis	labels
  modulo cucina vano sottostante con forno elettrico, due zone operative, con piastre tonde,	le zone cottura disponibili sono due	1
  modulo cucina profondità utile 70 cm, vano sottostante con forno elettrico, dispositivo a induzione, due moduli di cottura,	la cottura della cucina è a gas	0
  unità di cottura disposta su vano con ante, funziona a induzione, profondità utile 90 cm,	la cucina misura novante centimetri di profondità	1

• Loss: CrossEntropyLoss

Evaluation Dataset

json

• Dataset: json
• Size: 138,976 evaluation samples
• Columns: premises, hypothesis, and labels
• Approximate statistics based on the first 1000 samples:

  	premises	hypothesis	labels
  type	string	string	int
  details	min: 46 characters mean: 101.55 characters max: 154 characters	min: 12 characters mean: 33.34 characters max: 50 characters	0: ~45.90% 1: ~47.90% 2: ~6.20%

• Samples:

  premises	hypothesis	labels
  piano cottura due zone operative, con forno a gas,	la cucina ha un forno integrato	0
  unità di cottura a induzione, sistema con forno a gas, ampiezza di novanta centimetri, con quattro piastre cottura,	la cucina è profonda 90 cm	1
  modulo cucina funzionamento a induzione, dimensione teglie di gn1/1 due moduli di cottura, dotata di forno a gas,	la teglie del forno hanno dimensione gn1/1	1

• Loss: CrossEntropyLoss

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 32
• learning_rate: 1e-05
• num_train_epochs: 1
• warmup_steps: 46325
• bf16: True
• load_best_model_at_end: 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: 32
• per_device_eval_batch_size: 32
• 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: 1
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.0
• warmup_steps: 46325
• 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: 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}
• 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
0.0345	500	1.9008	1.5107
0.0691	1000	0.8958	0.6929
0.1036	1500	0.5841	0.4844
0.1382	2000	0.4403	0.3719
0.1727	2500	0.3578	0.2772
0.2072	3000	0.2732	0.2048
0.2418	3500	0.2117	0.1658
0.2763	4000	0.1717	0.1290
0.3108	4500	0.1444	0.1118
0.3454	5000	0.1283	0.1053
0.3799	5500	0.1136	0.1067
0.4145	6000	0.1066	0.0932
0.4490	6500	0.0987	0.0774
0.4835	7000	0.0864	0.0848
0.5181	7500	0.0849	0.0744
0.5526	8000	0.0796	0.0578
0.5871	8500	0.0671	0.0604
0.6217	9000	0.0656	0.0514
0.6562	9500	0.0609	0.0473

Framework Versions

• Python: 3.12.3
• Sentence Transformers: 5.1.1
• Transformers: 4.56.2
• PyTorch: 2.8.0+cu128
• Accelerate: 1.10.1
• Datasets: 4.1.1
• Tokenizers: 0.22.1

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",
}