juanpablomesa's picture
Add new SentenceTransformer model.
8a2a3f1 verified
metadata
base_model: BAAI/bge-small-en-v1.5
datasets: []
language:
  - en
library_name: sentence-transformers
license: apache-2.0
metrics:
  - cosine_accuracy@1
  - cosine_accuracy@3
  - cosine_accuracy@5
  - cosine_accuracy@10
  - cosine_precision@1
  - cosine_precision@3
  - cosine_precision@5
  - cosine_precision@10
  - cosine_recall@1
  - cosine_recall@3
  - cosine_recall@5
  - cosine_recall@10
  - cosine_ndcg@10
  - cosine_mrr@10
  - cosine_map@100
pipeline_tag: sentence-similarity
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:4012
  - loss:MultipleNegativesRankingLoss
widget:
  - source_sentence: >-
      Extensive messenger RNA editing generates transcript and protein diversity
      in genes involved in neural excitability, as previously described, as well
      as in genes participating in a broad range of other cellular functions. 
    sentences:
      - Do cephalopods use RNA editing less frequently than other species?
      - GV1001 vaccine targets which enzyme?
      - Which event results in the acetylation of S6K1?
  - source_sentence: >-
      Yes, exposure to household furry pets influences the gut microbiota of
      infants.
    sentences:
      - Can pets affect infant microbiomed?
      - What is the mode of action of Thiazovivin?
      - What are the effects of CAMK4 inhibition?
  - source_sentence: >-
      In children with heart failure evidence of the effect of enalapril is
      empirical. Enalapril was clinically safe and effective in 50% to 80% of
      for children with cardiac failure secondary to congenital heart
      malformations before and after cardiac surgery,  impaired ventricular
      function , valvar regurgitation,  congestive cardiomyopathy,  , arterial
      hypertension, life-threatening arrhythmias coexisting with circulatory
      insufficiency.   

      ACE inhibitors have shown a transient beneficial effect on heart failure
      due to anticancer drugs and possibly a beneficial effect in muscular
      dystrophy-associated cardiomyopathy, which deserves further studies.
    sentences:
      - Which receptors can be evaluated with the [18F]altanserin?
      - >-
        In what proportion of children with heart failure has Enalapril been
        shown to be safe and effective?
      - Which major signaling pathways are regulated by RIP1?
  - source_sentence: >-
      Cellular senescence-associated heterochromatic foci (SAHFS) are a novel
      type of chromatin condensation involving alterations of linker histone H1
      and linker DNA-binding proteins. SAHFS can be formed by a variety of cell
      types, but their mechanism of action remains unclear.
    sentences:
      - >-
        What is the relationship between the X chromosome and a  neutrophil
        drumstick?
      - Which microRNAs are involved in exercise adaptation?
      - How are SAHFS created?
  - source_sentence: >-
      Multicluster Pcdh diversity is required for mouse olfactory neural circuit
      assembly. The vertebrate clustered protocadherin (Pcdh) cell surface
      proteins are encoded by three closely linked gene clusters (Pcdhα, Pcdhβ,
      and Pcdhγ). Although deletion of individual Pcdh clusters had subtle
      phenotypic consequences, the loss of all three clusters (tricluster
      deletion) led to a severe axonal arborization defect and loss of
      self-avoidance.
    sentences:
      - >-
        What are the effects of the deletion of all three Pcdh clusters
        (tricluster deletion) in mice?
      - what is the role of MEF-2 in cardiomyocyte differentiation?
      - >-
        How many periods of regulatory innovation led to the evolution of
        vertebrates?
model-index:
  - name: BGE small finetuned BIOASQ
    results:
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: BAAI/bge small en v1.5
          type: BAAI/bge-small-en-v1.5
        metrics:
          - type: cosine_accuracy@1
            value: 0.8415841584158416
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.925035360678925
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.942008486562942
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.958981612446959
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.8415841584158416
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.30834512022630833
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.18840169731258838
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.09589816124469587
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.8415841584158416
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.925035360678925
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.942008486562942
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.958981612446959
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.9047357964584107
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.886919916481444
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.8877807671526188
            name: Cosine Map@100

BGE small finetuned BIOASQ

This is a sentence-transformers model finetuned from BAAI/bge-small-en-v1.5. 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: BAAI/bge-small-en-v1.5
  • Maximum Sequence Length: 512 tokens
  • Output Dimensionality: 384 tokens
  • Similarity Function: Cosine Similarity
  • Language: en
  • License: apache-2.0

Model Sources

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 512, 'do_lower_case': True}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 384, '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("juanpablomesa/bge-small-bioasq-1epochs-batch32")
# Run inference
sentences = [
    'Multicluster Pcdh diversity is required for mouse olfactory neural circuit assembly. The vertebrate clustered protocadherin (Pcdh) cell surface proteins are encoded by three closely linked gene clusters (Pcdhα, Pcdhβ, and Pcdhγ). Although deletion of individual Pcdh clusters had subtle phenotypic consequences, the loss of all three clusters (tricluster deletion) led to a severe axonal arborization defect and loss of self-avoidance.',
    'What are the effects of the deletion of all three Pcdh clusters (tricluster deletion) in mice?',
    'How many periods of regulatory innovation led to the evolution of vertebrates?',
]
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

Metric Value
cosine_accuracy@1 0.8416
cosine_accuracy@3 0.925
cosine_accuracy@5 0.942
cosine_accuracy@10 0.959
cosine_precision@1 0.8416
cosine_precision@3 0.3083
cosine_precision@5 0.1884
cosine_precision@10 0.0959
cosine_recall@1 0.8416
cosine_recall@3 0.925
cosine_recall@5 0.942
cosine_recall@10 0.959
cosine_ndcg@10 0.9047
cosine_mrr@10 0.8869
cosine_map@100 0.8878

Training Details

Training Dataset

Unnamed Dataset

  • Size: 4,012 training samples
  • Columns: positive and anchor
  • Approximate statistics based on the first 1000 samples:
    positive anchor
    type string string
    details
    • min: 3 tokens
    • mean: 63.38 tokens
    • max: 485 tokens
    • min: 5 tokens
    • mean: 16.13 tokens
    • max: 49 tokens
  • Samples:
    positive anchor
    Aberrant patterns of H3K4, H3K9, and H3K27 histone lysine methylation were shown to result in histone code alterations, which induce changes in gene expression, and affect the proliferation rate of cells in medulloblastoma. What is the implication of histone lysine methylation in medulloblastoma?
    STAG1/STAG2 proteins are tumour suppressor proteins that suppress cell proliferation and are essential for differentiation. What is the role of STAG1/STAG2 proteins in differentiation?
    The association between cell phone use and incident glioblastoma remains unclear. Some studies have reported that cell phone use was associated with incident glioblastoma, and with reduced survival of patients diagnosed with glioblastoma. However, other studies have repeatedly replicated to find an association between cell phone use and glioblastoma. What is the association between cell phone use and glioblastoma?
  • 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: 16
  • learning_rate: 2e-05
  • 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: 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: 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
  • batch_sampler: no_duplicates
  • multi_dataset_batch_sampler: proportional

Training Logs

Epoch Step Training Loss BAAI/bge-small-en-v1.5_cosine_map@100
0.0794 10 0.5344 -
0.1587 20 0.4615 -
0.2381 30 0.301 -
0.3175 40 0.2169 -
0.3968 50 0.1053 -
0.4762 60 0.1432 -
0.5556 70 0.1589 -
0.6349 80 0.1458 -
0.7143 90 0.1692 -
0.7937 100 0.1664 -
0.8730 110 0.1252 -
0.9524 120 0.1243 -
1.0 126 - 0.8858
0.0794 10 0.1393 -
0.1587 20 0.1504 -
0.2381 30 0.1009 -
0.3175 40 0.0689 -
0.3968 50 0.0301 -
0.4762 60 0.0647 -
0.5556 70 0.0748 -
0.6349 80 0.0679 -
0.7143 90 0.1091 -
0.7937 100 0.0953 -
0.8730 110 0.089 -
0.9524 120 0.0758 -
1.0 126 - 0.8878
0.0794 10 0.092 -
0.1587 20 0.0748 -
0.2381 30 0.0392 -
0.3175 40 0.014 -
0.3968 50 0.0057 -
0.4762 60 0.0208 -
0.5556 70 0.0173 -
0.6349 80 0.0195 -
0.7143 90 0.0349 -
0.7937 100 0.0483 -
0.8730 110 0.0254 -
0.9524 120 0.0325 -
1.0 126 - 0.8883
1.0317 130 0.0582 -
1.1111 140 0.0475 -
1.1905 150 0.0325 -
1.2698 160 0.0058 -
1.3492 170 0.0054 -
1.4286 180 0.0047 -
1.5079 190 0.0076 -
1.5873 200 0.0091 -
1.6667 210 0.0232 -
1.7460 220 0.0147 -
1.8254 230 0.0194 -
1.9048 240 0.0186 -
1.9841 250 0.0141 -
2.0 252 - 0.8857
2.0635 260 0.037 -
2.1429 270 0.0401 -
2.2222 280 0.0222 -
2.3016 290 0.0134 -
2.3810 300 0.008 -
2.4603 310 0.0199 -
2.5397 320 0.017 -
2.6190 330 0.0164 -
2.6984 340 0.0344 -
2.7778 350 0.0352 -
2.8571 360 0.0346 -
2.9365 370 0.0256 -
3.0 378 - 0.8868
0.7937 100 0.0064 0.8878

Framework Versions

  • Python: 3.11.5
  • Sentence Transformers: 3.0.1
  • Transformers: 4.41.2
  • PyTorch: 2.1.2+cu121
  • Accelerate: 0.31.0
  • Datasets: 2.19.1
  • 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",
}

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}
}