hyojuuun/gte_base_MIMICCXR_FT

SentenceTransformer based on thenlper/gte-base

This is a sentence-transformers model finetuned from thenlper/gte-base. 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: thenlper/gte-base
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 tokens
• 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: BertModel 
  (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})
  (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.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("hyojuuun/gte_base_MIMICCXR_FT")
# Run inference
sentences = [
    'The atient is status post coronary artery bypass graft surgery. The heart is mildly enlarged. There is a large hiatal hernia with an air-fluid level. Otherwise, the mediastinal and hilar contours are unremarkable. The lungs appear clear. The chest is hyperinflated. There is no pleural effusion or pneumothorax. Bony structures are unremarkable.  ',
    'No evidence of acute disease. Hyperinflation. Large hiatal hernia. Status post coronary artery bypass graft surgery. ',
    '1. Left apical pneumothorax still small, but considerably larger. Left base pneumothorax also slightly larger. 2. Minimal lucency adjacent to the the aortic knob may also represent part of the left lung pneumothorax. Attention to this area on followup films to exclude any mediastinal air is requested. 3. Extensive subcutaneous emphysema, equivocally slightly greater than on the prior film. 4. Minimal interval change in position of the left chest tube. 5. Right pneumothorax also increased, still small in width, but now seen not only at the right lung apex, but also along the right lateral chest wall and at the right costophrenic angle in the adjoining lung base. ',
]
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

Semantic Similarity

• Dataset: validation
• Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.8023
spearman_cosine	0.8105
pearson_manhattan	0.8243
spearman_manhattan	0.8105
pearson_euclidean	0.8245
spearman_euclidean	0.8105
pearson_dot	0.8023
spearman_dot	0.8105
pearson_max	0.8245
spearman_max	0.8105

Training Details

Training Dataset

Unnamed Dataset

• Size: 206,874 training samples
• Columns: sentence_0, sentence_1, and label
• Approximate statistics based on the first 1000 samples:

  	sentence_0	sentence_1	label
  type	string	string	float
  details	min: 3 tokens mean: 78.31 tokens max: 324 tokens	min: 4 tokens mean: 26.68 tokens max: 165 tokens	min: 0.0 mean: 0.5 max: 1.0

• Samples:

  sentence_0	sentence_1	label
  The lung volumes are low which accentuates the linear and interstitial opacities. An ill-defined opacity in the left lung in the third/fourth interspace has increased since the prior can be early pneumonia. No pneumothorax. Mild to moderate gastric and small bowel distension partially visualized.	No evidence of acute cardiopulmonary disease.	0.0
  PA and lateral views of the chest were provided demonstrating no focal consolidation, effusion or pneumothorax. The cardiomediastinal silhouette is normal. Bony structures are intact. No free air below the right hemidiaphragm.	No acute intrathoracic process.	1.0
  Previously seen right-sided PICC is no longer seen. Enlargement of the cardiomediastinal silhouette is grossly stable. There are low lung volumes, which accentuate the bronchovascular markings. No focal consolidation is seen. There is no pleural effusion or pneumothorax.	Low lung volumes but no focal consolidation to suggest pneumonia.	1.0

• Loss: ContrastiveLoss with these parameters:

  {
      "distance_metric": "SiameseDistanceMetric.COSINE_DISTANCE",
      "margin": 0.5,
      "size_average": true
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 96
• per_device_eval_batch_size: 96
• multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 96
• per_device_eval_batch_size: 96
• 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
• num_train_epochs: 3
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.0
• 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}
• 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: False
• hub_always_push: False
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• include_inputs_for_metrics: False
• 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
• eval_use_gather_object: False
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: round_robin

Training Logs

Epoch	Step	Training Loss	validation_spearman_max
0.0464	100	-	0.6178
0.0928	200	-	0.6904
0.1392	300	-	0.7290
0.1856	400	-	0.7596
0.2320	500	0.0191	0.7715
0.2784	600	-	0.7783
0.3248	700	-	0.7851
0.3712	800	-	0.7885
0.4176	900	-	0.7942
0.4640	1000	0.0118	0.7965
0.5104	1100	-	0.8061
0.5568	1200	-	0.8035
0.6032	1300	-	0.8082
0.6497	1400	-	0.8105

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.2.0
• Transformers: 4.44.2
• PyTorch: 2.4.1+cu121
• Accelerate: 0.34.2
• Datasets: 3.0.1
• Tokenizers: 0.19.1

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",
}

ContrastiveLoss

@inproceedings{hadsell2006dimensionality,
    author={Hadsell, R. and Chopra, S. and LeCun, Y.},
    booktitle={2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06)},
    title={Dimensionality Reduction by Learning an Invariant Mapping},
    year={2006},
    volume={2},
    number={},
    pages={1735-1742},
    doi={10.1109/CVPR.2006.100}
}