Steve77/bge-base-bible-retrieval

BGE base bible test

This is a sentence-transformers model finetuned from BAAI/bge-base-en-v1.5 on the json dataset. 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 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • json
• Language: fr
• 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("Steve77/bge-base-bible-retrieval")
# Run inference
sentences = [
    "Quand les Lévites devaient-ils se présenter pour louer et célébrer l'Éternel?",
    'Chaque matin et chaque soir.',
    "Cinq mille talents d'or et dix mille talents d'argent ont été donnés.",
]
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

• Datasets: dim_768, dim_512, dim_256, dim_128 and dim_64
• Evaluated with InformationRetrievalEvaluator

Metric	dim_768	dim_512	dim_256	dim_128	dim_64
cosine_accuracy@1	0.1336	0.1277	0.1249	0.1094	0.0894
cosine_accuracy@3	0.188	0.1855	0.1746	0.1569	0.1318
cosine_accuracy@5	0.2139	0.2057	0.1974	0.1785	0.1501
cosine_accuracy@10	0.251	0.2421	0.2327	0.2084	0.1769
cosine_precision@1	0.1336	0.1277	0.1249	0.1094	0.0894
cosine_precision@3	0.0627	0.0618	0.0582	0.0523	0.0439
cosine_precision@5	0.0428	0.0411	0.0395	0.0357	0.03
cosine_precision@10	0.0251	0.0242	0.0233	0.0208	0.0177
cosine_recall@1	0.1336	0.1277	0.1249	0.1094	0.0894
cosine_recall@3	0.188	0.1855	0.1746	0.1569	0.1318
cosine_recall@5	0.2139	0.2057	0.1974	0.1785	0.1501
cosine_recall@10	0.251	0.2421	0.2327	0.2084	0.1769
cosine_ndcg@10	0.1882	0.1815	0.1744	0.1557	0.1303
cosine_mrr@10	0.1686	0.1626	0.1563	0.1392	0.1158
cosine_map@100	0.174	0.168	0.1614	0.1441	0.1207

Training Details

Training Dataset

json

• Dataset: json
• Size: 47,560 training samples
• Columns: anchor and positive
• Approximate statistics based on the first 1000 samples:

  	anchor	positive
  type	string	string
  details	min: 8 tokens mean: 21.23 tokens max: 45 tokens	min: 3 tokens mean: 25.14 tokens max: 110 tokens

• Samples:

  anchor	positive
  Quels sont les noms des fils de Schobal?	Aljan, Manahath, Ébal, Schephi et Onam
  Quels sont les noms des fils de Tsibeon?	Ajja et Ana
  Qui est le fils d'Ana?	Dischon

• 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: 16
• per_device_eval_batch_size: 16
• gradient_accumulation_steps: 16
• learning_rate: 2e-05
• lr_scheduler_type: cosine
• warmup_ratio: 0.1
• bf16: True
• 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
• torch_empty_cache_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: 3
• 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: 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: 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
• eval_on_start: False
• use_liger_kernel: False
• eval_use_gather_object: False
• prompts: None
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	dim_768_cosine_ndcg@10	dim_512_cosine_ndcg@10	dim_256_cosine_ndcg@10	dim_128_cosine_ndcg@10	dim_64_cosine_ndcg@10
0.0538	10	12.8804	-	-	-	-	-
0.1076	20	12.4714	-	-	-	-	-
0.1615	30	11.8263	-	-	-	-	-
0.2153	40	11.014	-	-	-	-	-
0.2691	50	10.1609	-	-	-	-	-
0.3229	60	10.6807	-	-	-	-	-
0.3767	70	9.3215	-	-	-	-	-
0.4305	80	10.3719	-	-	-	-	-
0.4844	90	9.4147	-	-	-	-	-
0.5382	100	9.5567	-	-	-	-	-
0.5920	110	8.7699	-	-	-	-	-
0.6458	120	9.0428	-	-	-	-	-
0.6996	130	9.0977	-	-	-	-	-
0.7534	140	8.0843	-	-	-	-	-
0.8073	150	8.1363	-	-	-	-	-
0.8611	160	7.5306	-	-	-	-	-
0.9149	170	7.7972	-	-	-	-	-
0.9687	180	7.9644	-	-	-	-	-
0.9956	185	-	0.1917	0.1879	0.1784	0.1583	0.1268
1.0225	190	7.6124	-	-	-	-	-
1.0764	200	6.6315	-	-	-	-	-
1.1302	210	7.2313	-	-	-	-	-
1.1840	220	6.5394	-	-	-	-	-
1.2378	230	6.7843	-	-	-	-	-
1.2916	240	6.9276	-	-	-	-	-
1.3454	250	7.2281	-	-	-	-	-
1.3993	260	6.9158	-	-	-	-	-
1.4531	270	6.5158	-	-	-	-	-
1.5069	280	6.916	-	-	-	-	-
1.5607	290	6.5717	-	-	-	-	-
1.6145	300	6.9225	-	-	-	-	-
1.6683	310	7.3981	-	-	-	-	-
1.7222	320	6.894	-	-	-	-	-
1.7760	330	6.0293	-	-	-	-	-
1.8298	340	5.9389	-	-	-	-	-
1.8836	350	5.959	-	-	-	-	-
1.9374	360	6.4268	-	-	-	-	-
1.9913	370	6.7366	-	-	-	-	-
1.9966	371	-	0.2012	0.1965	0.1862	0.1633	0.1361
2.0451	380	5.7871	-	-	-	-	-
2.0989	390	5.7358	-	-	-	-	-
2.1527	400	6.0964	-	-	-	-	-
2.2065	410	5.8331	-	-	-	-	-
2.2603	420	5.6152	-	-	-	-	-
2.3142	430	6.5018	-	-	-	-	-
2.3680	440	5.9798	-	-	-	-	-
2.4218	450	6.0598	-	-	-	-	-
2.4756	460	5.8222	-	-	-	-	-
2.5294	470	6.303	-	-	-	-	-
2.5832	480	5.9648	-	-	-	-	-
2.6371	490	6.415	-	-	-	-	-
2.6909	500	7.084	-	-	-	-	-
2.7447	510	5.692	-	-	-	-	-
2.7985	520	5.7706	-	-	-	-	-
2.8523	530	5.6943	-	-	-	-	-
2.9062	540	5.6817	-	-	-	-	-
2.9600	550	6.1265	-	-	-	-	-
2.9869	555	-	0.1882	0.1815	0.1744	0.1557	0.1303

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.11.11
• Sentence Transformers: 3.3.1
• Transformers: 4.45.2
• PyTorch: 2.5.1
• Accelerate: 1.2.1
• Datasets: 2.19.1
• Tokenizers: 0.20.1

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}
}