wwydmanski/arctic-pubmed-v0.1

SentenceTransformer based on Snowflake/snowflake-arctic-embed-l-v2.0

This is a sentence-transformers model finetuned from Snowflake/snowflake-arctic-embed-l-v2.0 on the json dataset. It maps sentences & paragraphs to a 1024-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: Snowflake/snowflake-arctic-embed-l-v2.0
• Maximum Sequence Length: 8192 tokens
• Output Dimensionality: 1024 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': 8192, 'do_lower_case': False}) with Transformer model: PeftModelForFeatureExtraction 
  (1): Pooling({'word_embedding_dimension': 1024, '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 = [
    'Risk-based water quality monitoring framework',
    'Development of a new risk-based framework to guide investment in water quality monitoring. ',
    'Water quality monitoring strategies - A review and future perspectives. ',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]

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

Evaluation

Metrics

Triplet

• Dataset: triplet-dev
• Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.802

Training Details

Training Dataset

json

• Dataset: json
• Size: 10,053 training samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 4 tokens mean: 10.58 tokens max: 33 tokens	min: 5 tokens mean: 26.91 tokens max: 79 tokens	min: 4 tokens mean: 15.99 tokens max: 61 tokens

• Samples:

  anchor	positive	negative
  Pediatric Infectious Disease Control	[Urgent tasks in scientific studies concerning the control of infectious diseases in children].	Pediatric workforce: a look at pediatric infectious diseases data from the American Board of Pediatrics.
  Thermal coefficient of phase shift	Thermal characteristics of phase shift in jacketed optical fibers.	Thermal effects.
  Renal biomarkers in heart failure	Current and novel renal biomarkers in heart failure.	Cardiac biomarkers of heart failure in chronic kidney disease.

• 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: 128
• per_device_eval_batch_size: 128
• num_train_epochs: 1
• lr_scheduler_type: cosine_with_restarts
• warmup_ratio: 0.1
• bf16: 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: 128
• per_device_eval_batch_size: 128
• 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: cosine_with_restarts
• 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: 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: 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
• 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
• 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
• average_tokens_across_devices: False
• prompts: None
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	triplet-dev_cosine_accuracy
0	0	-	0.58
0.0127	1	1.677	-
0.0253	2	1.7295	-
0.0380	3	1.6713	-
0.0506	4	1.4761	-
0.0633	5	1.3731	-
0.0759	6	1.8333	-
0.0886	7	1.3218	-
0.1013	8	1.1539	-
0.1139	9	1.4003	-
0.1266	10	1.4514	-
0.1392	11	1.0803	-
0.1519	12	1.183	-
0.1646	13	0.9984	-
0.1772	14	1.2043	-
0.1899	15	1.1367	-
0.2025	16	1.1863	-
0.2152	17	1.0185	-
0.2278	18	0.9038	-
0.2405	19	0.8942	-
0.2532	20	1.0396	-
0.2658	21	1.1067	-
0.2785	22	1.0281	-
0.2911	23	1.1479	-
0.3038	24	1.2893	-
0.3165	25	1.0388	-
0.3291	26	1.1925	-
0.3418	27	0.9564	-
0.3544	28	0.8533	-
0.3671	29	0.9999	-
0.3797	30	1.126	-
0.3924	31	0.9898	-
0.4051	32	0.8786	-
0.4177	33	0.9878	-
0.4304	34	1.0988	-
0.4430	35	0.9721	-
0.4557	36	0.838	-
0.4684	37	0.9935	-
0.4810	38	1.1439	-
0.4937	39	0.7076	-
0.5063	40	1.0033	-
0.5190	41	1.0411	-
0.5316	42	0.8646	-
0.5443	43	0.8991	-
0.5570	44	0.6337	-
0.5696	45	1.0695	-
0.5823	46	0.9144	-
0.5949	47	0.9248	-
0.6076	48	0.7711	-
0.6203	49	1.0001	-
0.6329	50	1.0151	-
0.6456	51	1.06	-
0.6582	52	0.8105	-
0.6709	53	0.6892	-
0.6835	54	1.1341	-
0.6962	55	0.9726	-
0.7089	56	0.8783	-
0.7215	57	0.8084	-
0.7342	58	1.089	-
0.7468	59	0.8486	-
0.7595	60	0.8507	-
0.7722	61	0.9502	-
0.7848	62	0.8178	-
0.7975	63	1.0142	-
0.8101	64	0.9516	-
0.8228	65	0.9399	-
0.8354	66	0.7602	-
0.8481	67	0.8389	-
0.8608	68	0.9234	-
0.8734	69	0.9747	-
0.8861	70	1.1591	-
0.8987	71	1.0074	-
0.9114	72	0.8169	-
0.9241	73	0.9561	-
0.9367	74	0.9406	-
0.9494	75	0.9603	-
0.9620	76	0.8758	-
0.9747	77	0.8546	-
0.9873	78	0.7313	-
1.0	79	0.6568	0.802

Framework Versions

• Python: 3.12.3
• Sentence Transformers: 3.3.1
• Transformers: 4.48.0.dev0
• PyTorch: 2.5.1
• Accelerate: 1.2.1
• Datasets: 2.19.0
• 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",
}

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}
}