dat-ai/bge-base-for_text2sql

BGE base SQL Matryoshka

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: 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("dat-ai/bge-base-for_text2sql")
# Run inference
sentences = [
    '\n  Given  the Column informations, generate an SQL query for the following question:\n  Column: Nomination | Actors Name | Film Name | Director | Country\n  Question: What was the film Falling up nominated for?\n  SQL Query: SELECT Nomination FROM table WHERE Film Name = Falling Up\n  ',
    'What was the film Falling up nominated for?',
    'Who wrote an episode watched by 19.01 million US viewers?',
]
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	0.4676	0.4678	0.4675	0.4677	0.4678
cosine_accuracy@3	0.4697	0.4697	0.4697	0.4696	0.4696
cosine_accuracy@5	0.4697	0.4697	0.4697	0.4698	0.4696
cosine_accuracy@10	0.4697	0.4697	0.4698	0.4698	0.4697
cosine_precision@1	0.4676	0.4678	0.4675	0.4677	0.4678
cosine_precision@3	0.1566	0.1566	0.1566	0.1565	0.1565
cosine_precision@5	0.0939	0.0939	0.0939	0.094	0.0939
cosine_precision@10	0.047	0.047	0.047	0.047	0.047
cosine_recall@1	0.4676	0.4678	0.4675	0.4677	0.4678
cosine_recall@3	0.4697	0.4697	0.4697	0.4696	0.4696
cosine_recall@5	0.4697	0.4697	0.4697	0.4698	0.4696
cosine_recall@10	0.4697	0.4697	0.4698	0.4698	0.4697
cosine_ndcg@10	0.4689	0.469	0.4689	0.4689	0.469
cosine_mrr@10	0.4686	0.4687	0.4686	0.4687	0.4687
cosine_map@100	0.4686	0.4687	0.4686	0.4687	0.4687

Training Details

Training Dataset

json

• Dataset: json
• Size: 56,355 training samples
• Columns: context and question
• Approximate statistics based on the first 1000 samples:

  	context	question
  type	string	string
  details	min: 45 tokens mean: 72.61 tokens max: 196 tokens	min: 7 tokens mean: 15.41 tokens max: 36 tokens

• Samples:

  context	question
  Given the Column informations, generate an SQL query for the following question: Column: State/territory	Text/background colour
  Given the Column informations, generate an SQL query for the following question: Column: State/territory	Text/background colour
  Given the Column informations, generate an SQL query for the following question: Column: State/territory	Text/background colour

• Loss: MatryoshkaLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: epoch
• per_device_train_batch_size: 16
• gradient_accumulation_steps: 8
• learning_rate: 2e-05
• num_train_epochs: 4
• lr_scheduler_type: cosine
• warmup_ratio: 0.1
• fp16: 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: 8
• per_gpu_train_batch_size: None
• per_gpu_eval_batch_size: None
• gradient_accumulation_steps: 8
• 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: 4
• 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: 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
• prompts: None
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

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
0.0227	10	1.773	-	-	-	-	-
0.0454	20	1.3231	-	-	-	-	-
0.0681	30	0.713	-	-	-	-	-
0.0908	40	0.286	-	-	-	-	-
0.1135	50	0.1013	-	-	-	-	-
0.1362	60	0.0635	-	-	-	-	-
0.1590	70	0.0453	-	-	-	-	-
0.1817	80	0.041	-	-	-	-	-
0.2044	90	0.039	-	-	-	-	-
0.2271	100	0.027	-	-	-	-	-
0.2498	110	0.0193	-	-	-	-	-
0.2725	120	0.0167	-	-	-	-	-
0.2952	130	0.016	-	-	-	-	-
0.3179	140	0.0197	-	-	-	-	-
0.3406	150	0.0217	-	-	-	-	-
0.3633	160	0.0162	-	-	-	-	-
0.3860	170	0.012	-	-	-	-	-
0.4087	180	0.013	-	-	-	-	-
0.4315	190	0.0255	-	-	-	-	-
0.4542	200	0.0229	-	-	-	-	-
0.4769	210	0.0181	-	-	-	-	-
0.4996	220	0.0195	-	-	-	-	-
0.5223	230	0.0199	-	-	-	-	-
0.5450	240	0.0144	-	-	-	-	-
0.5677	250	0.0102	-	-	-	-	-
0.5904	260	0.0101	-	-	-	-	-
0.6131	270	0.0095	-	-	-	-	-
0.6358	280	0.0173	-	-	-	-	-
0.6585	290	0.01	-	-	-	-	-
0.6812	300	0.0129	-	-	-	-	-
0.7039	310	0.0177	-	-	-	-	-
0.7267	320	0.0106	-	-	-	-	-
0.7494	330	0.0146	-	-	-	-	-
0.7721	340	0.0185	-	-	-	-	-
0.7948	350	0.0203	-	-	-	-	-
0.8175	360	0.0146	-	-	-	-	-
0.8402	370	0.0072	-	-	-	-	-
0.8629	380	0.0102	-	-	-	-	-
0.8856	390	0.0075	-	-	-	-	-
0.9083	400	0.0064	-	-	-	-	-
0.9310	410	0.0163	-	-	-	-	-
0.9537	420	0.0069	-	-	-	-	-
0.9764	430	0.0072	-	-	-	-	-
0.9991	440	0.0147	0.4688	0.4689	0.4688	0.4689	0.4689
1.0219	450	0.0151	-	-	-	-	-
1.0446	460	0.0135	-	-	-	-	-
1.0673	470	0.0189	-	-	-	-	-
1.0900	480	0.0121	-	-	-	-	-
1.1127	490	0.0064	-	-	-	-	-
1.1354	500	0.0111	-	-	-	-	-
1.1581	510	0.0103	-	-	-	-	-
1.1808	520	0.0144	-	-	-	-	-
1.2035	530	0.0151	-	-	-	-	-
1.2262	540	0.0062	-	-	-	-	-
1.2489	550	0.0104	-	-	-	-	-
1.2716	560	0.0046	-	-	-	-	-
1.2944	570	0.0056	-	-	-	-	-
1.3171	580	0.0073	-	-	-	-	-
1.3398	590	0.007	-	-	-	-	-
1.3625	600	0.0074	-	-	-	-	-
1.3852	610	0.0057	-	-	-	-	-
1.4079	620	0.0052	-	-	-	-	-
1.4306	630	0.0114	-	-	-	-	-
1.4533	640	0.0075	-	-	-	-	-
1.4760	650	0.0116	-	-	-	-	-
1.4987	660	0.0092	-	-	-	-	-
1.5214	670	0.0137	-	-	-	-	-
1.5441	680	0.0066	-	-	-	-	-
1.5668	690	0.0042	-	-	-	-	-
1.5896	700	0.0036	-	-	-	-	-
1.6123	710	0.0039	-	-	-	-	-
1.6350	720	0.0065	-	-	-	-	-
1.6577	730	0.0051	-	-	-	-	-
1.6804	740	0.0054	-	-	-	-	-
1.7031	750	0.0086	-	-	-	-	-
1.7258	760	0.0062	-	-	-	-	-
1.7485	770	0.0071	-	-	-	-	-
1.7712	780	0.0108	-	-	-	-	-
1.7939	790	0.009	-	-	-	-	-
1.8166	800	0.0075	-	-	-	-	-
1.8393	810	0.0039	-	-	-	-	-
1.8620	820	0.0047	-	-	-	-	-
1.8848	830	0.0037	-	-	-	-	-
1.9075	840	0.0037	-	-	-	-	-
1.9302	850	0.0064	-	-	-	-	-
1.9529	860	0.0047	-	-	-	-	-
1.9756	870	0.0034	-	-	-	-	-
1.9983	880	0.0061	0.4689	0.4689	0.4689	0.4690	0.4690
2.0210	890	0.0096	-	-	-	-	-
2.0437	900	0.0071	-	-	-	-	-
2.0664	910	0.0101	-	-	-	-	-
2.0891	920	0.0054	-	-	-	-	-
2.1118	930	0.0039	-	-	-	-	-
2.1345	940	0.0074	-	-	-	-	-
2.1573	950	0.0044	-	-	-	-	-
2.1800	960	0.0088	-	-	-	-	-
2.2027	970	0.0096	-	-	-	-	-
2.2254	980	0.0057	-	-	-	-	-
2.2481	990	0.0063	-	-	-	-	-
2.2708	1000	0.0026	-	-	-	-	-
2.2935	1010	0.0032	-	-	-	-	-
2.3162	1020	0.0027	-	-	-	-	-
2.3389	1030	0.0041	-	-	-	-	-
2.3616	1040	0.0052	-	-	-	-	-
2.3843	1050	0.0035	-	-	-	-	-
2.4070	1060	0.0025	-	-	-	-	-
2.4297	1070	0.0059	-	-	-	-	-
2.4525	1080	0.0048	-	-	-	-	-
2.4752	1090	0.0064	-	-	-	-	-
2.4979	1100	0.0066	-	-	-	-	-
2.5206	1110	0.0078	-	-	-	-	-
2.5433	1120	0.0057	-	-	-	-	-
2.5660	1130	0.0026	-	-	-	-	-
2.5887	1140	0.0021	-	-	-	-	-
2.6114	1150	0.0021	-	-	-	-	-
2.6341	1160	0.0047	-	-	-	-	-
2.6568	1170	0.0034	-	-	-	-	-
2.6795	1180	0.0044	-	-	-	-	-
2.7022	1190	0.0058	-	-	-	-	-
2.7250	1200	0.0043	-	-	-	-	-
2.7477	1210	0.0056	-	-	-	-	-
2.7704	1220	0.0076	-	-	-	-	-
2.7931	1230	0.0063	-	-	-	-	-
2.8158	1240	0.0033	-	-	-	-	-
2.8385	1250	0.0025	-	-	-	-	-
2.8612	1260	0.0019	-	-	-	-	-
2.8839	1270	0.0052	-	-	-	-	-
2.9066	1280	0.0021	-	-	-	-	-
2.9293	1290	0.0041	-	-	-	-	-
2.9520	1300	0.0035	-	-	-	-	-
2.9747	1310	0.0044	-	-	-	-	-
2.9974	1320	0.0035	-	-	-	-	-
2.9997	1321	-	0.469	0.469	0.469	0.469	0.469
3.0202	1330	0.0062	-	-	-	-	-
3.0429	1340	0.0047	-	-	-	-	-
3.0656	1350	0.008	-	-	-	-	-
3.0883	1360	0.0033	-	-	-	-	-
3.1110	1370	0.0025	-	-	-	-	-
3.1337	1380	0.0069	-	-	-	-	-
3.1564	1390	0.0035	-	-	-	-	-
3.1791	1400	0.0085	-	-	-	-	-
3.2018	1410	0.007	-	-	-	-	-
3.2245	1420	0.007	-	-	-	-	-
3.2472	1430	0.0052	-	-	-	-	-
3.2699	1440	0.0019	-	-	-	-	-
3.2926	1450	0.0022	-	-	-	-	-
3.3154	1460	0.0019	-	-	-	-	-
3.3381	1470	0.0028	-	-	-	-	-
3.3608	1480	0.0042	-	-	-	-	-
3.3835	1490	0.0023	-	-	-	-	-
3.4062	1500	0.0024	-	-	-	-	-
3.4289	1510	0.0036	-	-	-	-	-
3.4516	1520	0.0038	-	-	-	-	-
3.4743	1530	0.0063	-	-	-	-	-
3.4970	1540	0.0044	-	-	-	-	-
3.5197	1550	0.0064	-	-	-	-	-
3.5424	1560	0.0053	-	-	-	-	-
3.5651	1570	0.0019	-	-	-	-	-
3.5879	1580	0.0019	-	-	-	-	-
3.6106	1590	0.0017	-	-	-	-	-
3.6333	1600	0.004	-	-	-	-	-
3.6560	1610	0.0026	-	-	-	-	-
3.6787	1620	0.0031	-	-	-	-	-
3.7014	1630	0.0043	-	-	-	-	-
3.7241	1640	0.0032	-	-	-	-	-
3.7468	1650	0.0041	-	-	-	-	-
3.7695	1660	0.0069	-	-	-	-	-
3.7922	1670	0.0063	-	-	-	-	-
3.8149	1680	0.0038	-	-	-	-	-
3.8376	1690	0.0024	-	-	-	-	-
3.8603	1700	0.0018	-	-	-	-	-
3.8831	1710	0.0034	-	-	-	-	-
3.9058	1720	0.0016	-	-	-	-	-
3.9285	1730	0.0026	-	-	-	-	-
3.9512	1740	0.0037	-	-	-	-	-
3.9739	1750	0.0024	-	-	-	-	-
3.9966	1760	0.0027	0.4689	0.4690	0.4689	0.4689	0.4690

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.14
• Sentence Transformers: 3.3.0
• Transformers: 4.41.2
• PyTorch: 2.1.2+cu121
• Accelerate: 0.34.2
• 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}
}