pankajrajdeo/BioForge-bioformer-16L-foundational

SentenceTransformer

This is a sentence-transformers model trained on the json dataset. 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
• Maximum Sequence Length: 256 tokens
• Output Dimensionality: 384 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • json

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})
)

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("pankajrajdeo/BioForge-bioformer-16L-foundational")
# Run inference
sentences = [
    'pisiform joint',
    'It is a joint between the pisiform and triquetrum.',
    'UBERON',
]
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

Information Retrieval

• Dataset: foundational_eval
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.8787
cosine_accuracy@3	0.9251
cosine_accuracy@5	0.937
cosine_accuracy@10	0.9474
cosine_precision@1	0.8787
cosine_precision@3	0.4516
cosine_precision@5	0.2983
cosine_precision@10	0.1614
cosine_recall@1	0.651
cosine_recall@3	0.8341
cosine_recall@5	0.8719
cosine_recall@10	0.9022
cosine_ndcg@10	0.8778
cosine_mrr@10	0.9037
cosine_map@100	0.8544

Training Details

Training Dataset

json

• Dataset: json
• Size: 3,997,120 training samples
• Columns: anchor, positive, source, and pair_type
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	source	pair_type
  type	string	string	string	string
  details	min: 3 tokens mean: 12.57 tokens max: 256 tokens	min: 3 tokens mean: 15.69 tokens max: 256 tokens	min: 3 tokens mean: 5.12 tokens max: 7 tokens	min: 3 tokens mean: 3.06 tokens max: 9 tokens

• Samples:

  anchor	positive	source	pair_type
  IM - Intramuscular sedation	Intramuscular sedation	SNOMED_CT	synonym
  Metergoline	Sodium channel protein type 2 subunit alpha	DrugBank	target
  Trichomycterus sp. MBML6210_4	unclassified Trichomycterus	NCBITAXON	hierarchy

• Loss: main.MultipleNegativesSymmetricMarginLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 512
• gradient_accumulation_steps: 4
• learning_rate: 1.2e-05
• lr_scheduler_type: cosine
• warmup_ratio: 0.05
• bf16: True
• dataloader_num_workers: 16
• load_best_model_at_end: True
• gradient_checkpointing: 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: 512
• per_device_eval_batch_size: 8
• per_gpu_train_batch_size: None
• per_gpu_eval_batch_size: None
• gradient_accumulation_steps: 4
• eval_accumulation_steps: None
• torch_empty_cache_steps: None
• learning_rate: 1.2e-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: 3
• max_steps: -1
• lr_scheduler_type: cosine
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.05
• 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: 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: 16
• 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}
• 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: True
• 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

Training Logs

Epoch	Step	Training Loss	foundational_eval_cosine_ndcg@10
0.0518	100	1.0162	-
0.1035	200	0.7522	-
0.1553	300	0.644	-
0.2070	400	0.5971	-
0.2588	500	0.5651	-
0.3105	600	0.5391	-
0.3297	637	-	0.8536
0.3623	700	0.5306	-
0.4140	800	0.5122	-
0.4658	900	0.5024	-
0.5175	1000	0.494	-
0.5693	1100	0.4907	-
0.6210	1200	0.48	-
0.6593	1274	-	0.8639
0.6728	1300	0.47	-
0.7245	1400	0.4657	-
0.7763	1500	0.4643	-
0.8281	1600	0.4573	-
0.8798	1700	0.4555	-
0.9316	1800	0.4537	-
0.9833	1900	0.4431	-
0.9890	1911	-	0.8693
1.0347	2000	0.4356	-
1.0864	2100	0.4299	-
1.1382	2200	0.4278	-
1.1899	2300	0.4307	-
1.2417	2400	0.4242	-
1.2934	2500	0.4279	-
1.3183	2548	-	0.8723
1.3452	2600	0.4185	-
1.3969	2700	0.42	-
1.4487	2800	0.4189	-
1.5005	2900	0.4183	-
1.5522	3000	0.4143	-
1.6040	3100	0.4147	-
1.6479	3185	-	0.8748
1.6557	3200	0.413	-
1.7075	3300	0.4107	-
1.7592	3400	0.4114	-
1.8110	3500	0.4111	-
1.8627	3600	0.4073	-
1.9145	3700	0.4093	-
1.9662	3800	0.4057	-
1.9776	3822	-	0.8766
2.0176	3900	0.3993	-
2.0693	4000	0.3996	-
2.1211	4100	0.3987	-
2.1729	4200	0.4012	-
2.2246	4300	0.3979	-
2.2764	4400	0.3977	-
2.3069	4459	-	0.8774
2.3281	4500	0.3981	-
2.3799	4600	0.394	-
2.4316	4700	0.3946	-
2.4834	4800	0.395	-
2.5351	4900	0.3971	-
2.5869	5000	0.3963	-
2.6366	5096	-	0.8776
2.6386	5100	0.396	-
2.6904	5200	0.3976	-
2.7421	5300	0.3963	-
2.7939	5400	0.3985	-
2.8456	5500	0.3968	-
2.8974	5600	0.3973	-
2.9492	5700	0.3981	-
2.9662	5733	-	0.8778

Framework Versions

• Python: 3.11.11
• Sentence Transformers: 3.4.1
• Transformers: 4.53.2
• PyTorch: 2.6.0+cu124
• Accelerate: 1.5.2
• Datasets: 3.2.0
• Tokenizers: 0.21.0

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",
}