MANMEET75/pubmedbert-base-embedding-Chatbot-Matryoshk

pubmedbert-base-embedding Chatbot Matryoshka

This is a sentence-transformers model finetuned from NeuML/pubmedbert-base-embeddings. 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: NeuML/pubmedbert-base-embeddings
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 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': 512, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 768, '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("MANMEET75/pubmedbert-base-embedding-Chatbot-Matryoshk")
# Run inference
sentences = [
    "I can understand and respond in multiple Indian regional languages. Feel free to communicate with me in the language you're most comfortable with.",
    'Bharti, what languages can you understand and respond to?',
    'Bharti, can you provide tips for effective online communication?',
]
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]

Evaluation

Metrics

Information Retrieval

• Dataset: dim_768
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.7674
cosine_accuracy@3	0.907
cosine_accuracy@5	0.9302
cosine_accuracy@10	0.9302
cosine_precision@1	0.7674
cosine_precision@3	0.3023
cosine_precision@5	0.186
cosine_precision@10	0.093
cosine_recall@1	0.7674
cosine_recall@3	0.907
cosine_recall@5	0.9302
cosine_recall@10	0.9302
cosine_ndcg@10	0.8564
cosine_mrr@10	0.8314
cosine_map@100	0.835

Information Retrieval

• Dataset: dim_512
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6977
cosine_accuracy@3	0.8837
cosine_accuracy@5	0.9302
cosine_accuracy@10	0.9302
cosine_precision@1	0.6977
cosine_precision@3	0.2946
cosine_precision@5	0.186
cosine_precision@10	0.093
cosine_recall@1	0.6977
cosine_recall@3	0.8837
cosine_recall@5	0.9302
cosine_recall@10	0.9302
cosine_ndcg@10	0.832
cosine_mrr@10	0.7984
cosine_map@100	0.8017

Information Retrieval

• Dataset: dim_256
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.7907
cosine_accuracy@3	0.8837
cosine_accuracy@5	0.907
cosine_accuracy@10	0.907
cosine_precision@1	0.7907
cosine_precision@3	0.2946
cosine_precision@5	0.1814
cosine_precision@10	0.0907
cosine_recall@1	0.7907
cosine_recall@3	0.8837
cosine_recall@5	0.907
cosine_recall@10	0.907
cosine_ndcg@10	0.8533
cosine_mrr@10	0.8353
cosine_map@100	0.8392

Information Retrieval

• Dataset: dim_128
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6744
cosine_accuracy@3	0.814
cosine_accuracy@5	0.8837
cosine_accuracy@10	0.907
cosine_precision@1	0.6744
cosine_precision@3	0.2713
cosine_precision@5	0.1767
cosine_precision@10	0.0907
cosine_recall@1	0.6744
cosine_recall@3	0.814
cosine_recall@5	0.8837
cosine_recall@10	0.907
cosine_ndcg@10	0.7942
cosine_mrr@10	0.7576
cosine_map@100	0.76

Information Retrieval

• Dataset: dim_64
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6047
cosine_accuracy@3	0.7442
cosine_accuracy@5	0.7907
cosine_accuracy@10	0.8605
cosine_precision@1	0.6047
cosine_precision@3	0.2481
cosine_precision@5	0.1581
cosine_precision@10	0.086
cosine_recall@1	0.6047
cosine_recall@3	0.7442
cosine_recall@5	0.7907
cosine_recall@10	0.8605
cosine_ndcg@10	0.722
cosine_mrr@10	0.6786
cosine_map@100	0.6823

Training Details

Training Dataset

Unnamed Dataset

• Size: 530 training samples
• Columns: positive and anchor
• Approximate statistics based on the first 1000 samples:

  	positive	anchor
  type	string	string
  details	min: 12 tokens mean: 36.83 tokens max: 107 tokens	min: 7 tokens mean: 18.54 tokens max: 30 tokens

• Samples:

  positive	anchor
  BharatPe Speaker comes with the following benefits: - Helps you avoid payment fraud - Lightweight & Easy installation process - Compatible with SIM & GPRS connectivity - Comes with a battery, no hassle of constant charging - Available in 10 Languages - Cashback Offers - Free replacement To Know more and place an order, tap below http://bharatpe.in/speaker.	What are the benefits of the BharatPe speaker?
  BharatPe Speaker comes with the following benefits: - Helps you avoid payment fraud - Lightweight & Easy installation process - Compatible with SIM & GPRS connectivity - Comes with a battery, no hassle of constant charging - Available in 10 Languages - Cashback Offers - Free replacement To Know more and place an order, tap below http://bharatpe.in/speaker.	What advantages does the BharatPe speaker offer?
  BharatPe Speaker comes with the following benefits: - Helps you avoid payment fraud - Lightweight & Easy installation process - Compatible with SIM & GPRS connectivity - Comes with a battery, no hassle of constant charging - Available in 10 Languages - Cashback Offers - Free replacement To Know more and place an order, tap below http://bharatpe.in/speaker.	Can you outline the benefits of using the BharatPe speaker?

• Loss: MatryoshkaLoss with these parameters:

  {
      "loss": "MultipleNegativesRankingLoss",
      "matryoshka_dims": [
          768,
          512,
          256,
          128,
          64
      ],
      "matryoshka_weights": [
          1,
          1,
          1,
          1,
          1
      ],
      "n_dims_per_step": -1
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: epoch
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 16
• gradient_accumulation_steps: 16
• learning_rate: 2e-05
• num_train_epochs: 10
• lr_scheduler_type: cosine
• warmup_ratio: 0.1
• tf32: False
• load_best_model_at_end: True
• optim: adamw_torch_fused
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: epoch
• 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: 16
• 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: 10
• max_steps: -1
• lr_scheduler_type: cosine
• 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: False
• fp16_opt_level: O1
• half_precision_backend: auto
• bf16_full_eval: False
• fp16_full_eval: False
• tf32: False
• 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: 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_fused
• 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	dim_128_cosine_map@100	dim_256_cosine_map@100	dim_512_cosine_map@100	dim_64_cosine_map@100	dim_768_cosine_map@100
0.9412	1	-	0.4829	0.5338	0.5921	0.3235	0.6100
1.8824	2	-	0.5767	0.6175	0.6588	0.4176	0.6793
2.8235	3	-	0.6337	0.6776	0.6979	0.5083	0.7263
3.7647	4	-	0.6588	0.7257	0.7297	0.5840	0.7612
4.7059	5	-	0.7049	0.7766	0.7643	0.6151	0.7902
5.6471	6	-	0.7374	0.8257	0.7890	0.6519	0.7956
6.5882	7	-	0.7573	0.8261	0.7912	0.6689	0.7978
7.5294	8	-	0.7590	0.8275	0.7958	0.6811	0.8233
8.4706	9	-	0.76	0.8392	0.7998	0.6823	0.8234
9.4118	10	4.944	0.7600	0.8392	0.8017	0.6823	0.8350

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.0.1
• Transformers: 4.41.2
• PyTorch: 2.1.2+cu121
• Accelerate: 0.32.1
• 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",
}

MatryoshkaLoss

@misc{kusupati2024matryoshka,
    title={Matryoshka Representation Learning}, 
    author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
    year={2024},
    eprint={2205.13147},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

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}
}