khaled-omar/distilroberta-ai-job-embeddings

SentenceTransformer based on medicalai/ClinicalBERT

This is a sentence-transformers model finetuned from medicalai/ClinicalBERT on the parquet dataset. 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: medicalai/ClinicalBERT
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • parquet

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: DistilBertModel 
  (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("khaled-omar/distilroberta-ai-job-embeddings")
# Run inference
sentences = [
    'irregular,period,few months,moderate,few months ago,none,weight,90. 000,height,163. 000,temperature,98. 600,pulse,82. 000,respiration,19. 000 bp,systolic,110. 000 bp,diastolic,70. 000,sp,o,2,:,99,cap,blood sugar,ja,und,ice,ec,abd,an,l,girth,head,chest,ch ida ch vitamin d deficiency,polycystic ovary syndrome,ch ida ch vitamin d deficiency,polycystic ovary syndrome,ch,ida ch vitamin d deficiency,polycystic ovary syndrome,ch,ida ch vitamin d deficiency,polycystic ovary syndrome,no,no family,no,no,nation,grade 11,grade 11,grade 11,grade 11,no,no,no,no,normal,normal,normal,normal,_ cvs,cv,normal,normal,irregular period,cns,cn,normal,mu,normal,normal,normal,normal,normal,normal,irregular period',
    'Irregular menstruation, unspecified',
    'Radial styloid tenosynovitis [de Quervain]',
]
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: ai-job-validation and ai-job-test
• Evaluated with TripletEvaluator

Metric	ai-job-validation	ai-job-test
cosine_accuracy	0.9429	0.9291

Training Details

Training Dataset

parquet

• Dataset: parquet
• Size: 7,999 training samples
• Columns: Entities, PosLongDesc, and NegLongDesc
• Approximate statistics based on the first 1000 samples:

  	Entities	PosLongDesc	NegLongDesc
  type	string	string	string
  details	min: 3 tokens mean: 155.39 tokens max: 485 tokens	min: 4 tokens mean: 10.62 tokens max: 32 tokens	min: 3 tokens mean: 10.35 tokens max: 31 tokens

• Samples:

  Entities	PosLongDesc	NegLongDesc
  it,chiness,since 3 months,it,chiness,since,3 months,weight,90. 100,height,178. 000,temperature,98. 060,pulse,84. 000,respiration,0. 000 bp,sy,sto,122. 000 bp,dia,69. 000,sp,o,:,99,cap,blood sugar,ja,undice,ec,abd,an,rth,nonsignificant,nonsignificant,nonsignifican,t,no family,nonsignificant family,nonsignificant family,nonsignificant,no relevant family history,yes,married, smoker, carpenter,married, smoker, carpenter social,married, smoker, carpenter social history,nonsignificant,nonsignificant,nonsignificant,it,chiness,3 months,treatment	Rash and other nonspecific skin eruption	Acute nasopharyngitis [common cold]
  amc,dubai,united arab emirates,uma,pa,gut,hari,val,electrocard,gram,pt,amc,sitting,coherent,w /,can,nula,bra,chia,vital,85,18,res,normal,sao,100,air time,17,: 51 : 34,bp,120 / 81,cap,<,2,sec,temperature,> 4 reacts,>,4,reacts,total,gcs,15,pain,0,blood glucose,102,car,accident,drug overdose,intentional	Epileptic seizures related to external causes, not intractable, without status epilepticus	COVID-19
  amc gate,dubai,united arab emirates,ssi,test,airports,dubai,concourse,ent assessment,throat,transported,endorsed,pulse :,77r,14,res,normal %,sao,2 :,100,air time,05 :,26,:,00,bp,118 / 69,cap,<,2,sec,temperature,36. 7,pupil,left,>,4,reacts,right,>,4,reacts,gcs,15,pain,2,blood glucose,96,car,accident,no,throatpain	Pain in throat	Encounter for observation for suspected exposure to other biological agents ruled out

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Evaluation Dataset

parquet

• Dataset: parquet
• Size: 999 evaluation samples
• Columns: Entities, PosLongDesc, and NegLongDesc
• Approximate statistics based on the first 999 samples:

  	Entities	PosLongDesc	NegLongDesc
  type	string	string	string
  details	min: 4 tokens mean: 154.58 tokens max: 470 tokens	min: 4 tokens mean: 10.61 tokens max: 31 tokens	min: 4 tokens mean: 10.12 tokens max: 35 tokens

• Samples:

  Entities	PosLongDesc	NegLongDesc
  it,chy,redness,3 days,both,it,ching,mild,moderate,3 days,weight,50. 200,height,143. 000,temperature,98. 240,pulse,78. 000,respiration,0. 000 bp,systolic,0. 000 bp,dia,sto,lic,0. 000,sp,o,2,:,99,cap,blood sugar,ja,undice,ec,abd,no past medical history,no past medical history,unknown family medical history,negative family,chronic disease,no diabetic mellitus,no hypertention,negative family,chronic disease,no diabetic mellitus,no hypertention,no,7 years and,7 months,7 years,7 months,no,removal,int,removal,int,red,it,chy,it,chy,redness,3 days	Acute atopic conjunctivitis, bilateral	Deficiency of other specified B group vitamins
  pi,mples,pustules,plus,minus,cyst,both side,of the face,too,it,ching,skin,4,pi,notice,increase,laser removal,facial,expose,sun,pust,cyst,it,weight,52,.,800,height,159. 000,temperature,98. 100,pulse,93. 000,res,0. 000 bp,sy,sto,99. 000 bp,sto,60. 000,sp,o,98,cap,blood sugar,ja,undice,ec,no,no,ro,course,ro,not,course,no diabetic mellitus,no,les,no diabetic,mellit,us,no,les,basic,nation,nation,13,years,months,15 years,11 months,old,pu,ules,plus,cyst,side	Local infection of the skin and subcutaneous tissue, unspecified	Inflammatory polyarthropathy
  respiratory rate,sp,pain,sy,lic,bp,mm,dia,bp,mm,height,weight,00 kg,repeat,prescription	Menopausal and female climacteric states	COVID-19

• 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: 16
• per_device_eval_batch_size: 16
• learning_rate: 2e-05
• num_train_epochs: 1
• warmup_ratio: 0.1
• 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: 16
• 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
• learning_rate: 2e-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: 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
• prompts: None
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	Validation Loss	ai-job-validation_cosine_accuracy	ai-job-test_cosine_accuracy
-1	-1	-	-	0.5495	-
0.2	100	2.8729	1.8172	0.8789	-
0.4	200	2.085	1.4398	0.9259	-
0.6	300	1.8233	1.3448	0.9339	-
0.8	400	1.6871	1.2579	0.9409	-
1.0	500	1.4881	1.2327	0.9429	-
-1	-1	-	-	0.9429	0.9291

Framework Versions

• Python: 3.11.4
• Sentence Transformers: 3.4.1
• Transformers: 4.41.2
• PyTorch: 2.3.1+cpu
• Accelerate: 1.3.0
• Datasets: 3.2.0
• 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",
}

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}
}