README.md

metadata

language: []
library_name: sentence-transformers
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - dataset_size:10K<n<100K
  - loss:MultipleNegativesSymmetricRankingLoss
base_model: distilbert/distilbert-base-uncased
metrics:
  - pearson_cosine
  - spearman_cosine
  - pearson_manhattan
  - spearman_manhattan
  - pearson_euclidean
  - spearman_euclidean
  - pearson_dot
  - spearman_dot
  - pearson_max
  - spearman_max
widget:
  - source_sentence: what is GOGO
    sentences:
      - What is Viasat
      - are we flying into Tel Aviv
      - how do i correct a name in term
  - source_sentence: What is EU 261
    sentences:
      - is puj a EU compensation country
      - can i take my bicycle on af
      - flight delays over 6 hours
  - source_sentence: Can i get wifi
    sentences:
      - which aircrafts do not have wifi
      - military traveling with pet
      - baggage delay to carousel
  - source_sentence: austin airport
    sentences:
      - What time is IAH open
      - amex card free checked bag
      - what is upgrade companion
  - source_sentence: pets in cargo
    sentences:
      - can a pet travel in cargo
      - baggage exceptions for Amex
      - how do I get sky priority
pipeline_tag: sentence-similarity
model-index:
  - name: SentenceTransformer based on distilbert/distilbert-base-uncased
    results:
      - task:
          type: semantic-similarity
          name: Semantic Similarity
        dataset:
          name: eval examples
          type: eval_examples
        metrics:
          - type: pearson_cosine
            value: .nan
            name: Pearson Cosine
          - type: spearman_cosine
            value: .nan
            name: Spearman Cosine
          - type: pearson_manhattan
            value: .nan
            name: Pearson Manhattan
          - type: spearman_manhattan
            value: .nan
            name: Spearman Manhattan
          - type: pearson_euclidean
            value: .nan
            name: Pearson Euclidean
          - type: spearman_euclidean
            value: .nan
            name: Spearman Euclidean
          - type: pearson_dot
            value: .nan
            name: Pearson Dot
          - type: spearman_dot
            value: .nan
            name: Spearman Dot
          - type: pearson_max
            value: .nan
            name: Pearson Max
          - type: spearman_max
            value: .nan
            name: Spearman Max

SentenceTransformer based on distilbert/distilbert-base-uncased

This is a sentence-transformers model finetuned from distilbert/distilbert-base-uncased. 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: distilbert/distilbert-base-uncased
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 tokens
• Similarity Function: Cosine Similarity

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: DistilBertModel 
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
)

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("pjbhaumik/biencoder-finetune-model-v9")
# Run inference
sentences = [
    'pets in cargo',
    'can a pet travel in cargo',
    'baggage exceptions for Amex',
]
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

Semantic Similarity

• Dataset: eval_examples
• Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	nan
spearman_cosine	nan
pearson_manhattan	nan
spearman_manhattan	nan
pearson_euclidean	nan
spearman_euclidean	nan
pearson_dot	nan
spearman_dot	nan
pearson_max	nan
spearman_max	nan

Training Details

Training Dataset

Unnamed Dataset

• Size: 15,488 training samples
• Columns: sentence_0, sentence_1, and label
• Approximate statistics based on the first 1000 samples:

  	sentence_0	sentence_1	label
  type	string	string	int
  details	min: 4 tokens mean: 10.4 tokens max: 47 tokens	min: 4 tokens mean: 10.14 tokens max: 37 tokens	1: 100.00%

• Samples:

  sentence_0	sentence_1	label
  how to use a companion certificate on delta.com	SHOPPING ON DELTA.COM FOR AMEX CERT	1
  is jamaica can be booked with companion certificate	what areas can the American Express companion certificate be applied to	1
  how do i book award travel on klm	can you book an air france ticket with miles	1

• Loss: MultipleNegativesSymmetricRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• num_train_epochs: 12
• multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• 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: 1
• eval_accumulation_steps: None
• learning_rate: 5e-05
• weight_decay: 0.0
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1
• num_train_epochs: 12
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.0
• 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: 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: False
• 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
• 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: batch_sampler
• multi_dataset_batch_sampler: round_robin

Training Logs

Click to expand

Epoch	Step	Training Loss	eval_examples_spearman_max
0.1033	100	-	nan
0.2066	200	-	nan
0.3099	300	-	nan
0.4132	400	-	nan
0.5165	500	0.7655	nan
0.6198	600	-	nan
0.7231	700	-	nan
0.8264	800	-	nan
0.9298	900	-	nan
1.0	968	-	nan
1.0331	1000	0.3727	nan
1.1364	1100	-	nan
1.2397	1200	-	nan
1.3430	1300	-	nan
1.4463	1400	-	nan
1.5496	1500	0.2686	nan
1.6529	1600	-	nan
1.7562	1700	-	nan
1.8595	1800	-	nan
1.9628	1900	-	nan
2.0	1936	-	nan
2.0661	2000	0.2709	nan
2.1694	2100	-	nan
2.2727	2200	-	nan
2.3760	2300	-	nan
2.4793	2400	-	nan
2.5826	2500	0.231	nan
2.6860	2600	-	nan
2.7893	2700	-	nan
2.8926	2800	-	nan
2.9959	2900	-	nan
3.0	2904	-	nan
3.0992	3000	0.2461	nan
3.2025	3100	-	nan
3.3058	3200	-	nan
3.4091	3300	-	nan
3.5124	3400	-	nan
3.6157	3500	0.2181	nan
3.7190	3600	-	nan
3.8223	3700	-	nan
3.9256	3800	-	nan
4.0	3872	-	nan
4.0289	3900	-	nan
4.1322	4000	0.2288	nan
4.2355	4100	-	nan
4.3388	4200	-	nan
4.4421	4300	-	nan
4.5455	4400	-	nan
4.6488	4500	0.2123	nan
4.7521	4600	-	nan
4.8554	4700	-	nan
4.9587	4800	-	nan
5.0	4840	-	nan
5.0620	4900	-	nan
5.1653	5000	0.2254	nan
5.2686	5100	-	nan
5.3719	5200	-	nan
5.4752	5300	-	nan
5.5785	5400	-	nan
5.6818	5500	0.2077	nan
5.7851	5600	-	nan
5.8884	5700	-	nan
5.9917	5800	-	nan
6.0	5808	-	nan
6.0950	5900	-	nan
6.1983	6000	0.218	nan
6.3017	6100	-	nan
6.4050	6200	-	nan
6.5083	6300	-	nan
6.6116	6400	-	nan
6.7149	6500	0.206	nan
6.8182	6600	-	nan
6.9215	6700	-	nan
7.0	6776	-	nan
7.0248	6800	-	nan
7.1281	6900	-	nan
7.2314	7000	0.2126	nan
7.3347	7100	-	nan
7.4380	7200	-	nan
7.5413	7300	-	nan
7.6446	7400	-	nan
7.7479	7500	0.2065	nan
7.8512	7600	-	nan
7.9545	7700	-	nan
8.0	7744	-	nan
8.0579	7800	-	nan
8.1612	7900	-	nan
8.2645	8000	0.2068	nan
8.3678	8100	-	nan
8.4711	8200	-	nan
8.5744	8300	-	nan
8.6777	8400	-	nan
8.7810	8500	0.2014	nan
8.8843	8600	-	nan
8.9876	8700	-	nan
9.0	8712	-	nan
9.0909	8800	-	nan
9.1942	8900	-	nan
9.2975	9000	0.2057	nan
9.4008	9100	-	nan
9.5041	9200	-	nan
9.6074	9300	-	nan
9.7107	9400	-	nan
9.8140	9500	0.1969	nan
9.9174	9600	-	nan
10.0	9680	-	nan
10.0207	9700	-	nan
10.1240	9800	-	nan
10.2273	9900	-	nan
10.3306	10000	0.2023	nan
10.4339	10100	-	nan
10.5372	10200	-	nan
10.6405	10300	-	nan
10.7438	10400	-	nan
10.8471	10500	0.1946	nan
10.9504	10600	-	nan
11.0	10648	-	nan
11.0537	10700	-	nan
11.1570	10800	-	nan
11.2603	10900	-	nan
11.3636	11000	0.1982	nan
11.4669	11100	-	nan
11.5702	11200	-	nan
11.6736	11300	-	nan
11.7769	11400	-	nan
11.8802	11500	0.1919	nan
11.9835	11600	-	nan
12.0	11616	-	nan

Framework Versions

• Python: 3.10.14
• Sentence Transformers: 3.0.0
• Transformers: 4.41.2
• PyTorch: 2.1.0
• 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",
}