ve88ifz2/snowflake-arctic-embed-m-klej-dyk-v0.1

snowflake-arctic-embed-m-klej-dyk

This is a sentence-transformers model finetuned from Snowflake/snowflake-arctic-embed-m. 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: Snowflake/snowflake-arctic-embed-m
• 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': 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 = [
    'Chłopiec z Nariokotome',
    'ile wynosiła objętość mózgu chłopca z Nariokotome?',
    'gdzie znajduje się czwarty polski cmentarz katyński?',
]
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	0.1851
cosine_accuracy@3	0.4808
cosine_accuracy@5	0.625
cosine_accuracy@10	0.726
cosine_precision@1	0.1851
cosine_precision@3	0.1603
cosine_precision@5	0.125
cosine_precision@10	0.0726
cosine_recall@1	0.1851
cosine_recall@3	0.4808
cosine_recall@5	0.625
cosine_recall@10	0.726
cosine_ndcg@10	0.4479
cosine_mrr@10	0.359
cosine_map@100	0.3672

Information Retrieval

• Dataset: dim_512
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1755
cosine_accuracy@3	0.4712
cosine_accuracy@5	0.613
cosine_accuracy@10	0.7019
cosine_precision@1	0.1755
cosine_precision@3	0.1571
cosine_precision@5	0.1226
cosine_precision@10	0.0702
cosine_recall@1	0.1755
cosine_recall@3	0.4712
cosine_recall@5	0.613
cosine_recall@10	0.7019
cosine_ndcg@10	0.4334
cosine_mrr@10	0.3474
cosine_map@100	0.3564

Information Retrieval

• Dataset: dim_256
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1562
cosine_accuracy@3	0.4543
cosine_accuracy@5	0.5649
cosine_accuracy@10	0.6731
cosine_precision@1	0.1562
cosine_precision@3	0.1514
cosine_precision@5	0.113
cosine_precision@10	0.0673
cosine_recall@1	0.1562
cosine_recall@3	0.4543
cosine_recall@5	0.5649
cosine_recall@10	0.6731
cosine_ndcg@10	0.4103
cosine_mrr@10	0.3261
cosine_map@100	0.3351

Information Retrieval

• Dataset: dim_128
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1635
cosine_accuracy@3	0.3918
cosine_accuracy@5	0.5072
cosine_accuracy@10	0.6058
cosine_precision@1	0.1635
cosine_precision@3	0.1306
cosine_precision@5	0.1014
cosine_precision@10	0.0606
cosine_recall@1	0.1635
cosine_recall@3	0.3918
cosine_recall@5	0.5072
cosine_recall@10	0.6058
cosine_ndcg@10	0.3758
cosine_mrr@10	0.3027
cosine_map@100	0.3117

Information Retrieval

• Dataset: dim_64
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.149
cosine_accuracy@3	0.3389
cosine_accuracy@5	0.4183
cosine_accuracy@10	0.4928
cosine_precision@1	0.149
cosine_precision@3	0.113
cosine_precision@5	0.0837
cosine_precision@10	0.0493
cosine_recall@1	0.149
cosine_recall@3	0.3389
cosine_recall@5	0.4183
cosine_recall@10	0.4928
cosine_ndcg@10	0.3178
cosine_mrr@10	0.2621
cosine_map@100	0.2704

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: 94.61 tokens max: 512 tokens	min: 10 tokens mean: 30.71 tokens max: 76 tokens

• Samples:

  positive	anchor
  Marsz Ochotników (chin.	kto jest kompozytorem chińskiego hymnu narodowego Marsz Ochotników?
  Wybrane przykłady: Święta Rodzina – Maryja z Dzieciątkiem na ręku, niekiedy obok niej stoi św. Józef Rodzina Marii – przedstawienie w którym pojawia się Święta Rodzina oraz postaci spokrewnione z Marią. Maria w połogu (Maria in puerperio) – leżąca na łożu Maria opiekuje się Dzieciątkiem Maria karmiąca (Maria lactans) – Maria karmiąca swą piersią Dzieciątko Orantka – kobieta modląca się z podniesionymi rękami (częsty motyw ikon wschodnich); Sacra Conversazione – Matka Boska tronująca z Dzieciątkiem, otoczona stojącymi postaciami świętych Pietà – opłakująca Jezusa, trzymając na kolanach jego ciało po śmierci na krzyżu; Hodegetria – ujęcie popiersia Maryi, trzymającej na rękach małego Jezusa, częsty motyw w ikonach Eleusa – formalnie podobne do przedstawienia Hodegetrii lecz Maryja policzkiem przytula się do policzka Jezusa Immaculata – Niepokalane Poczęcie Najświętszej Maryi Panny.	kto zamiast Maryi trzyma nowonarodzonego Jezusa w scenie Bożego Narodzenia przedstawionej na poliptyku z Marią i Dzieciątkiem Jezus?
  Pomnik Josepha von Eichendorffa w Brzeziu Pomnik Josepha von Eichendorffa – odtworzony w 2006 roku pomnik znanego niemieckiego poety epoki romantyzmu związanego z ziemią raciborską, Josepha von Eichendorffa.	po ilu latach odtworzono wysadzony w 1945 roku pomnik Josepha von Eichendorffa w Raciborzu-Brzeziu?

• 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: 5
• 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: 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: 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

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	9.3155	-	-	-	-	-
0.1368	2	9.1788	-	-	-	-	-
0.2051	3	8.8387	-	-	-	-	-
0.2735	4	8.2961	-	-	-	-	-
0.3419	5	8.0242	-	-	-	-	-
0.4103	6	7.2329	-	-	-	-	-
0.4786	7	5.4386	-	-	-	-	-
0.5470	8	6.1186	-	-	-	-	-
0.6154	9	4.9714	-	-	-	-	-
0.6838	10	5.1958	-	-	-	-	-
0.7521	11	5.1135	-	-	-	-	-
0.8205	12	4.6971	-	-	-	-	-
0.8889	13	4.5559	-	-	-	-	-
0.9573	14	3.9357	0.2842	0.3098	0.3191	0.2238	0.3209
1.0256	15	3.7916	-	-	-	-	-
1.0940	16	3.6393	-	-	-	-	-
1.1624	17	3.7733	-	-	-	-	-
1.2308	18	3.6974	-	-	-	-	-
1.2991	19	3.5964	-	-	-	-	-
1.3675	20	3.4118	-	-	-	-	-
1.4359	21	3.2022	-	-	-	-	-
1.5043	22	2.8133	-	-	-	-	-
1.5726	23	3.0871	-	-	-	-	-
1.6410	24	2.9559	-	-	-	-	-
1.7094	25	2.8192	-	-	-	-	-
1.7778	26	3.462	-	-	-	-	-
1.8462	27	3.1435	-	-	-	-	-
1.9145	28	2.8001	-	-	-	-	-
1.9829	29	2.5643	0.3134	0.3359	0.3563	0.2588	0.3671
2.0513	30	2.4295	-	-	-	-	-
2.1197	31	2.3892	-	-	-	-	-
2.1880	32	2.5228	-	-	-	-	-
2.2564	33	2.4906	-	-	-	-	-
2.3248	34	2.5358	-	-	-	-	-
2.3932	35	2.2806	-	-	-	-	-
2.4615	36	2.0083	-	-	-	-	-
2.5299	37	2.5088	-	-	-	-	-
2.5983	38	2.0628	-	-	-	-	-
2.6667	39	2.193	-	-	-	-	-
2.7350	40	2.4783	-	-	-	-	-
2.8034	41	2.382	-	-	-	-	-
2.8718	42	2.2017	-	-	-	-	-
2.9402	43	1.9739	0.3111	0.3392	0.3572	0.2657	0.3659
3.0085	44	2.0332	-	-	-	-	-
3.0769	45	1.9983	-	-	-	-	-
3.1453	46	1.8612	-	-	-	-	-
3.2137	47	1.9897	-	-	-	-	-
3.2821	48	2.2514	-	-	-	-	-
3.3504	49	2.0092	-	-	-	-	-
3.4188	50	1.7399	-	-	-	-	-
3.4872	51	1.5825	-	-	-	-	-
3.5556	52	2.1501	-	-	-	-	-
3.6239	53	1.4505	-	-	-	-	-
3.6923	54	1.8575	-	-	-	-	-
3.7607	55	2.3882	-	-	-	-	-
3.8291	56	2.1119	-	-	-	-	-
3.8974	57	1.8992	-	-	-	-	-
3.9658	58	1.8323	0.3117	0.3365	0.3558	0.2683	0.3670
4.0342	59	1.5938	-	-	-	-	-
4.1026	60	1.552	-	-	-	-	-
4.1709	61	1.907	-	-	-	-	-
4.2393	62	1.8304	-	-	-	-	-
4.3077	63	1.8775	-	-	-	-	-
4.3761	64	1.8654	-	-	-	-	-
4.4444	65	1.7944	-	-	-	-	-
4.5128	66	1.8335	-	-	-	-	-
4.5812	67	1.8823	-	-	-	-	-
4.6496	68	1.6479	-	-	-	-	-
4.7179	69	1.5771	-	-	-	-	-
4.7863	70	2.1911	0.3117	0.3351	0.3564	0.2704	0.3672

• 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}
}