cferreiragonz/bge-base-fastdds-questions-5b-epochs

Fine tuning poc1-5e

This is a sentence-transformers model finetuned from BAAI/bge-base-en-v1.5. 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: BAAI/bge-base-en-v1.5
• 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': True}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 768, '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("cferreiragonz/bge-base-fastdds-questions-5b-epochs")
# Run inference
sentences = [
    '* "AUTOMATIC_LIVELINESS_QOS": The service takes the responsibility for\n  renewing the leases at the required rates, as long as the local\n  process where the participant is running and the link connecting it\n  to remote participants exists, the entities within the remote\n  participant will be considered alive. This kind is suitable for\n  applications that only need to detect whether a remote application\n  is still running.',
    'What is the primary mechanism used by the service to ensure that a particular entity on the network remains considered "alive" when using the LivelinessQosPolicy with the "AUTOMATIC_ LIVELINESS_ QOS" kind?',
    'What is the purpose of loading an XML profiles file before creating entities in Fast DDS?',
]
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	0.3333
cosine_accuracy@3	0.4918
cosine_accuracy@5	0.5524
cosine_accuracy@10	0.6247
cosine_precision@1	0.3333
cosine_precision@3	0.1639
cosine_precision@5	0.1105
cosine_precision@10	0.0625
cosine_recall@1	0.3333
cosine_recall@3	0.4918
cosine_recall@5	0.5524
cosine_recall@10	0.6247
cosine_ndcg@10	0.472
cosine_mrr@10	0.4239
cosine_map@100	0.4312

Information Retrieval

• Dataset: dim_512
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.331
cosine_accuracy@3	0.4872
cosine_accuracy@5	0.5455
cosine_accuracy@10	0.62
cosine_precision@1	0.331
cosine_precision@3	0.1624
cosine_precision@5	0.1091
cosine_precision@10	0.062
cosine_recall@1	0.331
cosine_recall@3	0.4872
cosine_recall@5	0.5455
cosine_recall@10	0.62
cosine_ndcg@10	0.4662
cosine_mrr@10	0.4179
cosine_map@100	0.425

Information Retrieval

• Dataset: dim_256
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.31
cosine_accuracy@3	0.4732
cosine_accuracy@5	0.5431
cosine_accuracy@10	0.6084
cosine_precision@1	0.31
cosine_precision@3	0.1577
cosine_precision@5	0.1086
cosine_precision@10	0.0608
cosine_recall@1	0.31
cosine_recall@3	0.4732
cosine_recall@5	0.5431
cosine_recall@10	0.6084
cosine_ndcg@10	0.452
cosine_mrr@10	0.4023
cosine_map@100	0.4107

Information Retrieval

• Dataset: dim_128
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.303
cosine_accuracy@3	0.4639
cosine_accuracy@5	0.5268
cosine_accuracy@10	0.5967
cosine_precision@1	0.303
cosine_precision@3	0.1546
cosine_precision@5	0.1054
cosine_precision@10	0.0597
cosine_recall@1	0.303
cosine_recall@3	0.4639
cosine_recall@5	0.5268
cosine_recall@10	0.5967
cosine_ndcg@10	0.443
cosine_mrr@10	0.3944
cosine_map@100	0.4032

Information Retrieval

• Dataset: dim_64
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.2797
cosine_accuracy@3	0.4289
cosine_accuracy@5	0.4942
cosine_accuracy@10	0.5641
cosine_precision@1	0.2797
cosine_precision@3	0.143
cosine_precision@5	0.0988
cosine_precision@10	0.0564
cosine_recall@1	0.2797
cosine_recall@3	0.4289
cosine_recall@5	0.4942
cosine_recall@10	0.5641
cosine_ndcg@10	0.4175
cosine_mrr@10	0.3711
cosine_map@100	0.3801

Training Details

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: epoch
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• gradient_accumulation_steps: 16
• learning_rate: 2e-05
• num_train_epochs: 5
• lr_scheduler_type: cosine
• warmup_ratio: 0.1
• fp16: True
• 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: 16
• 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: 5
• 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: True
• 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
0.6639	10	5.0927	-	-	-	-	-
0.9959	15	-	0.3916	0.3898	0.4021	0.3546	0.4027
1.3278	20	3.3958	-	-	-	-	-
1.9917	30	2.6034	0.3893	0.4034	0.4163	0.3719	0.4222
2.6556	40	2.1012	-	-	-	-	-
2.9876	45	-	0.3975	0.4085	0.4240	0.3780	0.4291
3.3195	50	1.8189	-	-	-	-	-
3.9834	60	1.715	0.4029	0.411	0.4236	0.3794	0.4288
4.6473	70	1.6089	-	-	-	-	-
4.9793	75	-	0.4032	0.4107	0.4250	0.3801	0.4312

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.13
• Sentence Transformers: 3.0.1
• Transformers: 4.41.2
• PyTorch: 2.1.2
• Accelerate: 0.30.1
• 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}
}