kivanc/test

fine tune with openfda

This is a sentence-transformers model finetuned from sentence-transformers/all-MiniLM-L6-v2. It maps sentences & paragraphs to a 384-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: sentence-transformers/all-MiniLM-L6-v2
• Maximum Sequence Length: 256 tokens
• Output Dimensionality: 384 tokens
• 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': 256, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 384, '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("kivanc/test")
# Run inference
sentences = [
    'The pH range of TissueBlue 0.025% Solution is between 7.3 and 7.6.',
    'The pH range of the solution is 4.5 to 7.5.',
    'I.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Evaluation

Metrics

Triplet

• Dataset: dev-eva
• Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	1.0
dot_accuracy	0.0
manhattan_accuracy	1.0
euclidean_accuracy	1.0
max_accuracy	1.0

Triplet

• Dataset: test-eva
• Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	1.0
dot_accuracy	0.0
manhattan_accuracy	1.0
euclidean_accuracy	1.0
max_accuracy	1.0

Training Details

Training Dataset

Unnamed Dataset

• Size: 5,344 training samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 3 tokens mean: 41.73 tokens max: 255 tokens	min: 3 tokens mean: 42.27 tokens max: 256 tokens	min: 3 tokens mean: 24.11 tokens max: 157 tokens

• Samples:

  anchor	positive	negative
  Table 6– Physical Decay Chart for Technetium-99m, Half-Life 6.02 Hours Hours Fraction Remaining Hours Fraction Remaining 0* 1.000 7 0.447 1 0.891 8 0.398 2 0.794 9 0.355 3 0.708 10 0.316 4 0.631 11 0.282 5 0.562 12 0.251 6 0.501 *Calibration Time	Table 9 Physical Decay Chart of Technetium 99m Tc, Half Life: 6 Hours *Calibration Time Hours Fraction Remaining Hours Fraction Remaining 0 * 1.000 5 0.562 1 0.891 6 0.501 2 0.794 8 0.398 3 0.708 10 0.316 4 0.631 12 0.251	-Gently massage into affected areas.
  The compound has the empirical formula C 43 H 68 ClNO 11 and the molecular weight of 810.47.	Its molecular formula is C 17 H 16 ClNO⋅C 4 H 4 O 4 and its molecular weight is 401.84 (free base: 285.8).	Intravesical instillation for the treatment of interstitial cystitis.
  Adempas 1, 1.5, 2 and 2.5 mg tablets contain, in addition, ferric oxide yellow.	Adempas 2 and 2.5 mg tablets contain, in addition, ferric oxide red.	Higher temperatures lead to greater losses.

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Evaluation Dataset

Unnamed Dataset

• Size: 1,336 evaluation samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 3 tokens mean: 38.62 tokens max: 256 tokens	min: 3 tokens mean: 38.35 tokens max: 256 tokens	min: 3 tokens mean: 24.91 tokens max: 189 tokens

• Samples:

  anchor	positive	negative
  Sacituzumab govitecan-hziy contains on average 7 to 8 molecules of SN-38 per antibody molecule.	An average of 2.3 molecules of SG3249 are attached to each antibody molecule.	Over this time period, blood pressure returns gradually to pretreatment levels.
  11 DESCRIPTION INVOKANA ® (canagliflozin) contains canagliflozin, an inhibitor of SGLT2, the transporter responsible for reabsorbing the majority of glucose filtered by the kidney.	Canagliflozin, the active ingredient of INVOKANA, is chemically known as (1 S )-1,5-anhydro-1-[3-[[5-(4-fluorophenyl)-2-thienyl]methyl]-4-methylphenyl]-D-glucitol hemihydrate and its molecular formula and weight are C 24 H 25 FO 5 S∙1/2 H 2 O and 453.53, respectively.	1 Evaluated Nuclear Structure Data File of the Oak Ridge Nuclear Data Project DOE (1985).
  Its molecular formula is C 25 H 37 NO 4 .	Its molecular formula is C 27 H 41 NO 8 .	GOOD LENS CARE PRACTICES: ☞ Always wash and rinse your hands before you handle your lenses.

• 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
• num_train_epochs: 10
• 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: 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
• 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: 10
• 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: 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
• use_liger_kernel: False
• eval_use_gather_object: False
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	loss	dev-eva_max_accuracy	test-eva_max_accuracy
0	0	-	-	1.0	-
0.2994	100	0.0288	0.0080	1.0	-
0.5988	200	0.0297	0.0089	1.0	-
0.8982	300	0.0283	0.0103	1.0	-
1.1976	400	0.0021	0.0111	1.0	-
1.4970	500	0.0008	0.0137	1.0	-
1.7964	600	0.0	0.0137	1.0	-
1.2754	700	0.0198	0.0109	1.0	-
1.5749	800	0.0239	0.0165	1.0	-
1.8743	900	0.0118	0.0133	1.0	-
2.1737	1000	0.0012	0.0117	1.0	-
2.4731	1100	0.0001	0.0116	1.0	-
2.7725	1200	0.0	0.0116	1.0	-
2.2515	1300	0.0041	0.0120	1.0	-
2.5509	1400	0.0063	0.0102	1.0	-
2.8503	1500	0.0039	0.0154	1.0	-
3.1497	1600	0.0008	0.0113	1.0	-
3.4491	1700	0.0	0.0110	1.0	-
3.7485	1800	0.0	0.0110	1.0	-
3.2275	1900	0.0017	0.0122	1.0	-
3.5269	2000	0.0023	0.0119	1.0	-
3.8263	2100	0.0019	0.0123	1.0	-
4.1257	2200	0.0006	0.0125	1.0	-
4.4251	2300	0.0	0.0120	1.0	-
4.7246	2400	0.0	0.0120	1.0	-
4.2036	2500	0.0009	0.0125	1.0	-
4.5030	2600	0.0012	0.0115	1.0	-
4.8024	2700	0.0013	0.0125	1.0	-
5.1018	2800	0.0004	0.0120	1.0	-
5.4012	2900	0.0	0.0118	1.0	-
5.7006	3000	0.0	0.0118	1.0	-
5.1796	3100	0.0006	0.0120	1.0	-
5.4790	3200	0.001	0.0118	1.0	-
5.7784	3300	0.001	0.0118	1.0	-
5.8982	3340	-	-	-	1.0

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.1.1
• Transformers: 4.45.2
• PyTorch: 2.5.1+cu121
• Accelerate: 1.1.1
• Datasets: 3.1.0
• Tokenizers: 0.20.3

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}
}