ve88ifz2/bge-base-en-v1.5-klej-dyk-v0.2

bge-base-en-v1.5-klej-dyk

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("sentence_transformers_model_id")
# Run inference
sentences = [
    'Herkules na rozstajach',
    'jak zinterpretować wymowę obrazu Herkules na rozstajach?',
    'Dowódcą grupy był Wiaczesław Razumowicz ps. „Chmara”.',
]
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	0.1731
cosine_accuracy@3	0.4615
cosine_accuracy@5	0.6226
cosine_accuracy@10	0.7356
cosine_precision@1	0.1731
cosine_precision@3	0.1538
cosine_precision@5	0.1245
cosine_precision@10	0.0736
cosine_recall@1	0.1731
cosine_recall@3	0.4615
cosine_recall@5	0.6226
cosine_recall@10	0.7356
cosine_ndcg@10	0.4434
cosine_mrr@10	0.3505
cosine_map@100	0.3574

Information Retrieval

• Dataset: dim_512
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1683
cosine_accuracy@3	0.4519
cosine_accuracy@5	0.601
cosine_accuracy@10	0.7091
cosine_precision@1	0.1683
cosine_precision@3	0.1506
cosine_precision@5	0.1202
cosine_precision@10	0.0709
cosine_recall@1	0.1683
cosine_recall@3	0.4519
cosine_recall@5	0.601
cosine_recall@10	0.7091
cosine_ndcg@10	0.4296
cosine_mrr@10	0.3406
cosine_map@100	0.3485

Information Retrieval

• Dataset: dim_256
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1923
cosine_accuracy@3	0.4543
cosine_accuracy@5	0.5913
cosine_accuracy@10	0.6899
cosine_precision@1	0.1923
cosine_precision@3	0.1514
cosine_precision@5	0.1183
cosine_precision@10	0.069
cosine_recall@1	0.1923
cosine_recall@3	0.4543
cosine_recall@5	0.5913
cosine_recall@10	0.6899
cosine_ndcg@10	0.4311
cosine_mrr@10	0.3488
cosine_map@100	0.3561

Information Retrieval

• Dataset: dim_128
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1635
cosine_accuracy@3	0.4159
cosine_accuracy@5	0.5168
cosine_accuracy@10	0.5986
cosine_precision@1	0.1635
cosine_precision@3	0.1386
cosine_precision@5	0.1034
cosine_precision@10	0.0599
cosine_recall@1	0.1635
cosine_recall@3	0.4159
cosine_recall@5	0.5168
cosine_recall@10	0.5986
cosine_ndcg@10	0.3764
cosine_mrr@10	0.3052
cosine_map@100	0.3152

Information Retrieval

• Dataset: dim_64
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1659
cosine_accuracy@3	0.351
cosine_accuracy@5	0.4399
cosine_accuracy@10	0.5288
cosine_precision@1	0.1659
cosine_precision@3	0.117
cosine_precision@5	0.088
cosine_precision@10	0.0529
cosine_recall@1	0.1659
cosine_recall@3	0.351
cosine_recall@5	0.4399
cosine_recall@10	0.5288
cosine_ndcg@10	0.3382
cosine_mrr@10	0.278
cosine_map@100	0.2877

Training Details

Training Dataset

Unnamed Dataset

• Size: 3,738 training samples
• Columns: positive and anchor
• Approximate statistics based on the first 1000 samples:

  	positive	anchor
  type	string	string
  details	min: 6 tokens mean: 90.01 tokens max: 512 tokens	min: 10 tokens mean: 30.82 tokens max: 76 tokens

• Samples:

  positive	anchor
  Londyńska premiera w Ambassadors Theatre na londyńskim West Endzie miała miejsce 25 listopada 1952 roku, a przedstawione grane jest do dziś (od 1974 r.) w sąsiednim St Martin's Theatre. W Polsce była wystawiana m.in. w Teatrze Nowym w Zabrzu.	w którym londyńskim muzeum wystawiana była instalacja My Bed?
  Theridion grallator osiąga długość 5 mm. U niektórych postaci na żółtym odwłoku występuje wzór przypominający uśmiechniętą lub śmiejącą się twarz klowna.	które pająki noszą na grzbiecie wzór przypominający uśmiechniętego klauna?
  W 1998 w wyniku sporów o wytyczenie granicy między dwoma państwami wybuchła wojna erytrejsko-etiopska. Zakończyła się porozumieniem zawartym w Algierze 12 grudnia 2000. Od tego czasu strefa graniczna jest patrolowana przez siły pokojowe ONZ.	jakie były skutki wojny erytrejsko-etiopskiej?

• 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
• num_train_epochs: 10
• lr_scheduler_type: cosine
• warmup_ratio: 0.1
• bf16: True
• tf32: 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
• 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: True
• fp16: False
• fp16_opt_level: O1
• half_precision_backend: auto
• bf16_full_eval: False
• fp16_full_eval: False
• tf32: True
• 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

Click to expand

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
0.0684	1	7.2706	-	-	-	-	-
0.1368	2	8.2776	-	-	-	-	-
0.2051	3	7.1399	-	-	-	-	-
0.2735	4	6.6905	-	-	-	-	-
0.3419	5	6.735	-	-	-	-	-
0.4103	6	7.0537	-	-	-	-	-
0.4786	7	6.871	-	-	-	-	-
0.5470	8	6.7277	-	-	-	-	-
0.6154	9	5.9853	-	-	-	-	-
0.6838	10	6.0518	-	-	-	-	-
0.7521	11	5.8291	-	-	-	-	-
0.8205	12	5.0064	-	-	-	-	-
0.8889	13	4.8572	-	-	-	-	-
0.9573	14	5.1899	0.2812	0.3335	0.3486	0.2115	0.3639
1.0256	15	4.2996	-	-	-	-	-
1.0940	16	4.1475	-	-	-	-	-
1.1624	17	4.6174	-	-	-	-	-
1.2308	18	4.394	-	-	-	-	-
1.2991	19	4.0255	-	-	-	-	-
1.3675	20	3.9722	-	-	-	-	-
1.4359	21	3.9509	-	-	-	-	-
1.5043	22	3.7674	-	-	-	-	-
1.5726	23	3.7572	-	-	-	-	-
1.6410	24	3.9463	-	-	-	-	-
1.7094	25	3.7151	-	-	-	-	-
1.7778	26	3.7771	-	-	-	-	-
1.8462	27	3.5228	-	-	-	-	-
1.9145	28	2.7906	-	-	-	-	-
1.9829	29	3.4555	0.3164	0.3529	0.3641	0.2636	0.3681
2.0513	30	2.737	-	-	-	-	-
2.1197	31	3.1976	-	-	-	-	-
2.1880	32	3.1363	-	-	-	-	-
2.2564	33	2.9706	-	-	-	-	-
2.3248	34	2.9629	-	-	-	-	-
2.3932	35	2.7226	-	-	-	-	-
2.4615	36	2.4378	-	-	-	-	-
2.5299	37	2.7201	-	-	-	-	-
2.5983	38	2.6802	-	-	-	-	-
2.6667	39	3.1613	-	-	-	-	-
2.7350	40	2.9344	-	-	-	-	-
2.8034	41	2.5254	-	-	-	-	-
2.8718	42	2.5617	-	-	-	-	-
2.9402	43	2.459	0.3197	0.3571	0.3640	0.2739	0.3733
3.0085	44	2.3785	-	-	-	-	-
3.0769	45	1.9408	-	-	-	-	-
3.1453	46	2.7095	-	-	-	-	-
3.2137	47	2.4774	-	-	-	-	-
3.2821	48	2.2178	-	-	-	-	-
3.3504	49	2.0884	-	-	-	-	-
3.4188	50	2.1044	-	-	-	-	-
3.4872	51	2.1504	-	-	-	-	-
3.5556	52	2.1177	-	-	-	-	-
3.6239	53	2.2283	-	-	-	-	-
3.6923	54	2.3964	-	-	-	-	-
3.7607	55	2.0972	-	-	-	-	-
3.8291	56	2.0961	-	-	-	-	-
3.8974	57	1.783	-	-	-	-	-
3.9658	58	2.1031	0.3246	0.3533	0.3603	0.2829	0.3687
4.0342	59	1.6699	-	-	-	-	-
4.1026	60	1.6675	-	-	-	-	-
4.1709	61	2.1672	-	-	-	-	-
4.2393	62	1.8881	-	-	-	-	-
4.3077	63	1.701	-	-	-	-	-
4.3761	64	1.9154	-	-	-	-	-
4.4444	65	1.4549	-	-	-	-	-
4.5128	66	1.5444	-	-	-	-	-
4.5812	67	1.8352	-	-	-	-	-
4.6496	68	1.7908	-	-	-	-	-
4.7179	69	1.6876	-	-	-	-	-
4.7863	70	1.7366	-	-	-	-	-
4.8547	71	1.8689	-	-	-	-	-
4.9231	72	1.4676	-	-	-	-	-
4.9915	73	1.5045	0.3170	0.3538	0.3606	0.2829	0.3675
5.0598	74	1.2155	-	-	-	-	-
5.1282	75	1.4365	-	-	-	-	-
5.1966	76	1.7451	-	-	-	-	-
5.2650	77	1.4537	-	-	-	-	-
5.3333	78	1.3813	-	-	-	-	-
5.4017	79	1.4035	-	-	-	-	-
5.4701	80	1.3912	-	-	-	-	-
5.5385	81	1.3286	-	-	-	-	-
5.6068	82	1.5153	-	-	-	-	-
5.6752	83	1.6745	-	-	-	-	-
5.7436	84	1.4323	-	-	-	-	-
5.8120	85	1.5299	-	-	-	-	-
5.8803	86	1.488	-	-	-	-	-
5.9487	87	1.5195	0.3206	0.3556	0.3530	0.2878	0.3605
6.0171	88	1.2999	-	-	-	-	-
6.0855	89	1.1511	-	-	-	-	-
6.1538	90	1.552	-	-	-	-	-
6.2222	91	1.35	-	-	-	-	-
6.2906	92	1.218	-	-	-	-	-
6.3590	93	1.1712	-	-	-	-	-
6.4274	94	1.3381	-	-	-	-	-
6.4957	95	1.1716	-	-	-	-	-
6.5641	96	1.2117	-	-	-	-	-
6.6325	97	1.5349	-	-	-	-	-
6.7009	98	1.4564	-	-	-	-	-
6.7692	99	1.3541	-	-	-	-	-
6.8376	100	1.2468	-	-	-	-	-
6.9060	101	1.1519	-	-	-	-	-
6.9744	102	1.2421	0.3150	0.3555	0.3501	0.2858	0.3575
7.0427	103	1.0096	-	-	-	-	-
7.1111	104	1.1405	-	-	-	-	-
7.1795	105	1.2958	-	-	-	-	-
7.2479	106	1.35	-	-	-	-	-
7.3162	107	1.1291	-	-	-	-	-
7.3846	108	0.9968	-	-	-	-	-
7.4530	109	1.0454	-	-	-	-	-
7.5214	110	1.102	-	-	-	-	-
7.5897	111	1.1328	-	-	-	-	-
7.6581	112	1.5988	-	-	-	-	-
7.7265	113	1.2992	-	-	-	-	-
7.7949	114	1.2572	-	-	-	-	-
7.8632	115	1.1414	-	-	-	-	-
7.9316	116	1.1432	-	-	-	-	-
8.0	117	1.1181	0.3154	0.3545	0.3509	0.2884	0.3578
8.0684	118	0.9365	-	-	-	-	-
8.1368	119	1.3286	-	-	-	-	-
8.2051	120	1.3711	-	-	-	-	-
8.2735	121	1.2001	-	-	-	-	-
8.3419	122	1.165	-	-	-	-	-
8.4103	123	1.0575	-	-	-	-	-
8.4786	124	1.105	-	-	-	-	-
8.5470	125	1.077	-	-	-	-	-
8.6154	126	1.2217	-	-	-	-	-
8.6838	127	1.3254	-	-	-	-	-
8.7521	128	1.2165	-	-	-	-	-
8.8205	129	1.3021	-	-	-	-	-
8.8889	130	1.0927	-	-	-	-	-
8.9573	131	1.3961	0.3150	0.3540	0.3490	0.2882	0.3588
9.0256	132	1.0779	-	-	-	-	-
9.0940	133	0.901	-	-	-	-	-
9.1624	134	1.313	-	-	-	-	-
9.2308	135	1.1409	-	-	-	-	-
9.2991	136	1.1635	-	-	-	-	-
9.3675	137	1.0244	-	-	-	-	-
9.4359	138	1.0576	-	-	-	-	-
9.5043	139	1.0101	-	-	-	-	-
9.5726	140	1.1516	0.3152	0.3561	0.3485	0.2877	0.3574

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.12.2
• Sentence Transformers: 3.0.0
• Transformers: 4.41.2
• PyTorch: 2.3.1
• Accelerate: 0.27.2
• 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}
}