arulpm/gte-ipb-clinical

GTE Multilingual fine-tuned on clinical-to-drug mapping

This is a sentence-transformers model finetuned from Alibaba-NLP/gte-multilingual-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: Alibaba-NLP/gte-multilingual-base
• Maximum Sequence Length: 8192 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity
• Language: multilingual
• 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': 8192, 'do_lower_case': False}) with Transformer model: NewModel 
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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("sentence_transformers_model_id")
# Run inference
sentences = [
    'Keluhan: nan\nAnamnesa Pemeriksaan Dokter: mual pusing',
    'Obat: Buscopan\nDeskripsi Obat: Obat untuk mengurangi kejang otot polos saluran pencernaan',
    'Obat: Blocand 16 mg\nDeskripsi Obat: Obat antihipertensi golongan ARB dosis tinggi untuk menurunkan tekanan darah',
]
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]

Training Details

Training Dataset

Unnamed Dataset

• Size: 217,680 training samples
• Columns: anchor and positive
• Approximate statistics based on the first 1000 samples:

  	anchor	positive
  type	string	string
  details	min: 17 tokens mean: 42.44 tokens max: 177 tokens	min: 15 tokens mean: 22.55 tokens max: 38 tokens

• Samples:

  anchor	positive
  Keluhan: nan Anamnesa Pemeriksaan Dokter: gusi kiri bengkak
  nyeri tenggorokan, ada amandel 2 hari yang lalu
  puisng	Obat: Eflagen 50 mg Deskripsi Obat: Obat antiinflamasi nonsteroid (NSAID) untuk mengurangi nyeri
  Keluhan: pasien mengatakan kuku kaki jempol kiri merah dan nyeri sejak 1 bulan yll hilang timbul Anamnesa Pemeriksaan Dokter: pasien mengatakan kuku kaki jempol kiri merah dan nyeri sejak 1 bulan yll hilang timbul. merah + bengkak +	Obat: Mefinal 500 mg Deskripsi Obat: Obat antiinflamasi nonsteroid (NSAID) untuk mengurangi nyeri
  Keluhan: batuk dahak sudah 1 minggu Anamnesa Pemeriksaan Dokter: batuk dahak sejak 1 minggu
  RPO : Acetylcystein, OBH	Obat: Cefixime 200 mg Deskripsi Obat: Antibiotik golongan sefalosporin untuk mengobati infeksi bakteri

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Evaluation Dataset

Unnamed Dataset

• Size: 24,187 evaluation samples
• Columns: anchor and positive
• Approximate statistics based on the first 1000 samples:

  	anchor	positive
  type	string	string
  details	min: 16 tokens mean: 42.79 tokens max: 112 tokens	min: 15 tokens mean: 22.79 tokens max: 38 tokens

• Samples:

  anchor	positive
  Keluhan: NYERI PERUT, MUAL, MUNTAH 1KALI HARI INI, Anamnesa Pemeriksaan Dokter: NYERI PERUT, MUAL, MUNTAH 1KALI HARI INI, tenggorokan gatla
  rpo myalanta
  alergi obat dsiangkal	Obat: Lambucid Deskripsi Obat: Antasida untuk meredakan gejala asam lambung berlebih
  Keluhan: badan lemas tadi pagi, mual- muntah- batuk pilek- Anamnesa Pemeriksaan Dokter: lemas sejak pagi ini. demam - pingsan - sempat terjatuh karena lemas. pola makan tidak teratur.
  rpo tablet tambah darah. riw anemia + dikatakan saat SMA 3 tahun yll. tidak ingat hb berapa.	Obat: Caviplex Deskripsi Obat: Suplemen multivitamin untuk memenuhi kebutuhan vitamin dan mineral
  Keluhan: rencana rujukan rsMD spP tgl 23-04-24
  keluhan hari ini kepala pusing terasa berat, batuk, nyeri badan. Anamnesa Pemeriksaan Dokter: rencana rujukan rsMD spP tgl 23-04-24 keluhan hari ini kepala pusing terasa berat, batuk, nyeri badan.	Obat: Profat sirup Deskripsi Obat: Suplemen penambah darah sirup untuk anak

• 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: 32
• per_device_eval_batch_size: 32
• num_train_epochs: 1
• warmup_ratio: 0.1
• fp16: 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: 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: 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: 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
• 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: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss	Validation Loss
0.0073	50	3.3431	-
0.0147	100	3.1904	-
0.0220	150	3.0541	-
0.0294	200	2.972	-
0.0367	250	2.8877	-
0.0441	300	2.8234	-
0.0514	350	2.749	-
0.0588	400	2.7435	-
0.0661	450	2.7368	-
0.0735	500	2.6943	-
0.0808	550	2.7168	-
0.0882	600	2.7194	-
0.0955	650	2.6096	-
0.1029	700	2.7118	-
0.1102	750	2.7036	-
0.1176	800	2.6625	-
0.1249	850	2.6362	-
0.1323	900	2.599	-
0.1396	950	2.572	-
0.1470	1000	2.6124	2.0072
0.1543	1050	2.5467	-
0.1617	1100	2.5713	-
0.1690	1150	2.5741	-
0.1764	1200	2.5794	-
0.1837	1250	2.5231	-
0.1911	1300	2.5312	-
0.1984	1350	2.4483	-
0.2058	1400	2.5178	-
0.2131	1450	2.4795	-
0.2205	1500	2.5426	-
0.2278	1550	2.502	-
0.2352	1600	2.5378	-
0.2425	1650	2.4746	-
0.2499	1700	2.4356	-
0.2572	1750	2.5303	-
0.2646	1800	2.514	-
0.2719	1850	2.5207	-
0.2793	1900	2.4671	-
0.2866	1950	2.4367	-
0.2940	2000	2.4873	1.9339
0.3013	2050	2.4513	-
0.3087	2100	2.4695	-
0.3160	2150	2.4309	-
0.3234	2200	2.4439	-
0.3307	2250	2.4242	-
0.3381	2300	2.4569	-
0.3454	2350	2.4157	-
0.3528	2400	2.4709	-
0.3601	2450	2.4202	-
0.3675	2500	2.4401	-
0.3748	2550	2.4096	-
0.3822	2600	2.3878	-
0.3895	2650	2.4766	-
0.3969	2700	2.4149	-
0.4042	2750	2.4197	-
0.4116	2800	2.3656	-
0.4189	2850	2.4679	-
0.4263	2900	2.3749	-
0.4336	2950	2.4146	-
0.4410	3000	2.3942	1.8871
0.4483	3050	2.418	-
0.4557	3100	2.4504	-
0.4630	3150	2.3759	-
0.4704	3200	2.3671	-
0.4777	3250	2.4433	-
0.4851	3300	2.4036	-
0.4924	3350	2.3539	-
0.4998	3400	2.3806	-
0.5071	3450	2.3737	-
0.5145	3500	2.4127	-
0.5218	3550	2.4243	-
0.5292	3600	2.3528	-
0.5365	3650	2.3788	-
0.5439	3700	2.3968	-
0.5512	3750	2.3896	-
0.5586	3800	2.3966	-
0.5659	3850	2.3571	-
0.5733	3900	2.3437	-
0.5806	3950	2.3353	-
0.5880	4000	2.3335	1.8599
0.5953	4050	2.3778	-
0.6027	4100	2.3929	-
0.6100	4150	2.3818	-
0.6174	4200	2.3874	-
0.6247	4250	2.3224	-
0.6321	4300	2.3317	-
0.6394	4350	2.3761	-
0.6468	4400	2.4066	-
0.6541	4450	2.3406	-
0.6615	4500	2.3844	-
0.6688	4550	2.2993	-
0.6762	4600	2.337	-
0.6835	4650	2.37	-
0.6909	4700	2.3126	-
0.6982	4750	2.3818	-
0.7056	4800	2.3849	-
0.7129	4850	2.3379	-
0.7203	4900	2.3518	-
0.7276	4950	2.3354	-
0.7350	5000	2.3443	1.8349
0.7423	5050	2.3396	-
0.7497	5100	2.3086	-
0.7570	5150	2.3392	-
0.7644	5200	2.3316	-
0.7717	5250	2.3092	-
0.7791	5300	2.3794	-
0.7864	5350	2.331	-
0.7938	5400	2.2554	-
0.8011	5450	2.3266	-
0.8085	5500	2.3314	-
0.8158	5550	2.3357	-
0.8232	5600	2.3523	-
0.8305	5650	2.3253	-
0.8379	5700	2.3021	-
0.8452	5750	2.3342	-
0.8526	5800	2.2839	-
0.8599	5850	2.3136	-
0.8673	5900	2.3562	-
0.8746	5950	2.2878	-
0.8820	6000	2.3219	1.8173
0.8893	6050	2.2941	-
0.8967	6100	2.3245	-
0.9040	6150	2.2561	-
0.9114	6200	2.3327	-
0.9187	6250	2.3047	-
0.9261	6300	2.2916	-
0.9334	6350	2.3495	-
0.9408	6400	1.9273	-
0.9481	6450	1.3917	-
0.9555	6500	1.4726	-
0.9628	6550	1.3922	-
0.9702	6600	1.4664	-
0.9775	6650	1.4329	-
0.9849	6700	1.4046	-
0.9922	6750	1.3891	-
0.9996	6800	1.4731	-

Framework Versions

• Python: 3.11.13
• Sentence Transformers: 4.1.0
• Transformers: 4.53.2
• PyTorch: 2.6.0+cu124
• Accelerate: 1.8.1
• Datasets: 2.14.4
• Tokenizers: 0.21.2

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}
}