metadata
base_model: llmrails/ember-v1
datasets: []
language: []
library_name: sentence-transformers
metrics:
- pearson_cosine
- spearman_cosine
- pearson_manhattan
- spearman_manhattan
- pearson_euclidean
- spearman_euclidean
- pearson_dot
- spearman_dot
- pearson_max
- spearman_max
pipeline_tag: sentence-similarity
tags:
- sentence-transformers
- sentence-similarity
- feature-extraction
- generated_from_trainer
- dataset_size:17500
- loss:ContrastiveLoss
widget:
- source_sentence: 260 Mount Prospect Apt A4
sentences:
- 254 Mount Apt 304
- '110 Nightin - Gale #10'
- 3100 35 Apt 2
- source_sentence: '20 Harding #2'
sentences:
- '1208 Barclay #2'
- '65 Chestnut # 72'
- '20 Harding # 2'
- source_sentence: 396 Manila Apt 2B
sentences:
- 108 Gaston 1
- '175 2nd #710'
- '1 - 02 Virginia #102B'
- source_sentence: '210 Gordon Fl #1'
sentences:
- 450 Raritan Ste C
- '19 Edsall #1'
- '210 Gordon # 1'
- source_sentence: '148 1 / 2 Mill #B'
sentences:
- '7918 Pershing #'
- '91 5th #1'
- 148 1 / 2 Mill Apt B
model-index:
- name: SentenceTransformer based on llmrails/ember-v1
results:
- task:
type: semantic-similarity
name: Semantic Similarity
dataset:
name: test
type: test
metrics:
- type: pearson_cosine
value: 0.7022202945624949
name: Pearson Cosine
- type: spearman_cosine
value: 0.5521115900667813
name: Spearman Cosine
- type: pearson_manhattan
value: 0.5902198760799219
name: Pearson Manhattan
- type: spearman_manhattan
value: 0.5601831188247873
name: Spearman Manhattan
- type: pearson_euclidean
value: 0.5899864734850421
name: Pearson Euclidean
- type: spearman_euclidean
value: 0.5598213668477258
name: Spearman Euclidean
- type: pearson_dot
value: 0.5811885721377421
name: Pearson Dot
- type: spearman_dot
value: 0.45745466821334696
name: Spearman Dot
- type: pearson_max
value: 0.7022202945624949
name: Pearson Max
- type: spearman_max
value: 0.5601831188247873
name: Spearman Max
- task:
type: semantic-similarity
name: Semantic Similarity
dataset:
name: validation
type: validation
metrics:
- type: pearson_cosine
value: 0.9438614062035536
name: Pearson Cosine
- type: spearman_cosine
value: 0.6566095423715015
name: Spearman Cosine
- type: pearson_manhattan
value: 0.9661648909940331
name: Pearson Manhattan
- type: spearman_manhattan
value: 0.6545897461863388
name: Spearman Manhattan
- type: pearson_euclidean
value: 0.9662831349240031
name: Pearson Euclidean
- type: spearman_euclidean
value: 0.6545799233453334
name: Spearman Euclidean
- type: pearson_dot
value: 0.7832494697144132
name: Pearson Dot
- type: spearman_dot
value: 0.6240051940767
name: Spearman Dot
- type: pearson_max
value: 0.9662831349240031
name: Pearson Max
- type: spearman_max
value: 0.6566095423715015
name: Spearman Max
SentenceTransformer based on llmrails/ember-v1
This is a sentence-transformers model finetuned from llmrails/ember-v1. It maps sentences & paragraphs to a 1024-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: llmrails/ember-v1
- Maximum Sequence Length: 512 tokens
- Output Dimensionality: 1024 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': True}) with Transformer model: BertModel
(1): Pooling({'word_embedding_dimension': 1024, '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})
)
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("jarredparrett/fine-tuned-address-model-ember-v1")
# Run inference
sentences = [
'148 1 / 2 Mill #B',
'148 1 / 2 Mill Apt B',
'7918 Pershing #',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]
# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]
Evaluation
Metrics
Semantic Similarity
- Dataset:
test
- Evaluated with
EmbeddingSimilarityEvaluator
Metric | Value |
---|---|
pearson_cosine | 0.7022 |
spearman_cosine | 0.5521 |
pearson_manhattan | 0.5902 |
spearman_manhattan | 0.5602 |
pearson_euclidean | 0.59 |
spearman_euclidean | 0.5598 |
pearson_dot | 0.5812 |
spearman_dot | 0.4575 |
pearson_max | 0.7022 |
spearman_max | 0.5602 |
Semantic Similarity
- Dataset:
validation
- Evaluated with
EmbeddingSimilarityEvaluator
Metric | Value |
---|---|
pearson_cosine | 0.9439 |
spearman_cosine | 0.6566 |
pearson_manhattan | 0.9662 |
spearman_manhattan | 0.6546 |
pearson_euclidean | 0.9663 |
spearman_euclidean | 0.6546 |
pearson_dot | 0.7832 |
spearman_dot | 0.624 |
pearson_max | 0.9663 |
spearman_max | 0.6566 |
Training Details
Training Dataset
Unnamed Dataset
- Size: 17,500 training samples
- Columns:
sentence_0
,sentence_1
, andlabel
- Approximate statistics based on the first 1000 samples:
sentence_0 sentence_1 label type string string int details - min: 5 tokens
- mean: 6.97 tokens
- max: 12 tokens
- min: 5 tokens
- mean: 6.96 tokens
- max: 12 tokens
- 0: ~17.80%
- 1: ~82.20%
- Samples:
sentence_0 sentence_1 label 94 Liberty 1
94 Liberty Flr 1
1
166 Randolph Apt 1
166 Randolph Flr 1
1
400 Dutch Apt E12
400 Dutch Neck Apt E12
1
- Loss:
ContrastiveLoss
with these parameters:{ "distance_metric": "SiameseDistanceMetric.COSINE_DISTANCE", "margin": 0.5, "size_average": true }
Training Hyperparameters
Non-Default Hyperparameters
eval_strategy
: stepsper_device_train_batch_size
: 16per_device_eval_batch_size
: 16num_train_epochs
: 4multi_dataset_batch_sampler
: round_robin
All Hyperparameters
Click to expand
overwrite_output_dir
: Falsedo_predict
: Falseeval_strategy
: stepsprediction_loss_only
: Trueper_device_train_batch_size
: 16per_device_eval_batch_size
: 16per_gpu_train_batch_size
: Noneper_gpu_eval_batch_size
: Nonegradient_accumulation_steps
: 1eval_accumulation_steps
: Nonetorch_empty_cache_steps
: Nonelearning_rate
: 5e-05weight_decay
: 0.0adam_beta1
: 0.9adam_beta2
: 0.999adam_epsilon
: 1e-08max_grad_norm
: 1num_train_epochs
: 4max_steps
: -1lr_scheduler_type
: linearlr_scheduler_kwargs
: {}warmup_ratio
: 0.0warmup_steps
: 0log_level
: passivelog_level_replica
: warninglog_on_each_node
: Truelogging_nan_inf_filter
: Truesave_safetensors
: Truesave_on_each_node
: Falsesave_only_model
: Falserestore_callback_states_from_checkpoint
: Falseno_cuda
: Falseuse_cpu
: Falseuse_mps_device
: Falseseed
: 42data_seed
: Nonejit_mode_eval
: Falseuse_ipex
: Falsebf16
: Falsefp16
: Falsefp16_opt_level
: O1half_precision_backend
: autobf16_full_eval
: Falsefp16_full_eval
: Falsetf32
: Nonelocal_rank
: 0ddp_backend
: Nonetpu_num_cores
: Nonetpu_metrics_debug
: Falsedebug
: []dataloader_drop_last
: Falsedataloader_num_workers
: 0dataloader_prefetch_factor
: Nonepast_index
: -1disable_tqdm
: Falseremove_unused_columns
: Truelabel_names
: Noneload_best_model_at_end
: Falseignore_data_skip
: Falsefsdp
: []fsdp_min_num_params
: 0fsdp_config
: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}fsdp_transformer_layer_cls_to_wrap
: Noneaccelerator_config
: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}deepspeed
: Nonelabel_smoothing_factor
: 0.0optim
: adamw_torchoptim_args
: Noneadafactor
: Falsegroup_by_length
: Falselength_column_name
: lengthddp_find_unused_parameters
: Noneddp_bucket_cap_mb
: Noneddp_broadcast_buffers
: Falsedataloader_pin_memory
: Truedataloader_persistent_workers
: Falseskip_memory_metrics
: Trueuse_legacy_prediction_loop
: Falsepush_to_hub
: Falseresume_from_checkpoint
: Nonehub_model_id
: Nonehub_strategy
: every_savehub_private_repo
: Falsehub_always_push
: Falsegradient_checkpointing
: Falsegradient_checkpointing_kwargs
: Noneinclude_inputs_for_metrics
: Falseeval_do_concat_batches
: Truefp16_backend
: autopush_to_hub_model_id
: Nonepush_to_hub_organization
: Nonemp_parameters
:auto_find_batch_size
: Falsefull_determinism
: Falsetorchdynamo
: Noneray_scope
: lastddp_timeout
: 1800torch_compile
: Falsetorch_compile_backend
: Nonetorch_compile_mode
: Nonedispatch_batches
: Nonesplit_batches
: Noneinclude_tokens_per_second
: Falseinclude_num_input_tokens_seen
: Falseneftune_noise_alpha
: Noneoptim_target_modules
: Nonebatch_eval_metrics
: Falseeval_on_start
: Falseeval_use_gather_object
: Falsebatch_sampler
: batch_samplermulti_dataset_batch_sampler
: round_robin
Training Logs
Epoch | Step | Training Loss | test_spearman_max | validation_spearman_max |
---|---|---|---|---|
0 | 0 | - | 0.5602 | - |
0.0914 | 100 | - | - | 0.6565 |
0.1828 | 200 | - | - | 0.6567 |
0.2742 | 300 | - | - | 0.6558 |
0.3656 | 400 | - | - | 0.6560 |
0.4570 | 500 | 0.0039 | - | 0.6560 |
0.5484 | 600 | - | - | 0.6555 |
0.6399 | 700 | - | - | 0.6559 |
0.7313 | 800 | - | - | 0.6561 |
0.8227 | 900 | - | - | 0.6555 |
0.9141 | 1000 | 0.0019 | - | 0.6558 |
1.0 | 1094 | - | - | 0.6560 |
1.0055 | 1100 | - | - | 0.6561 |
1.0969 | 1200 | - | - | 0.6560 |
1.1883 | 1300 | - | - | 0.6559 |
1.2797 | 1400 | - | - | 0.6555 |
1.3711 | 1500 | 0.0014 | - | 0.6558 |
1.4625 | 1600 | - | - | 0.6560 |
1.5539 | 1700 | - | - | 0.6557 |
1.6453 | 1800 | - | - | 0.6561 |
1.7367 | 1900 | - | - | 0.6561 |
1.8282 | 2000 | 0.001 | - | 0.6562 |
1.9196 | 2100 | - | - | 0.6563 |
2.0 | 2188 | - | - | 0.6564 |
2.0110 | 2200 | - | - | 0.6565 |
2.1024 | 2300 | - | - | 0.6565 |
2.1938 | 2400 | - | - | 0.6560 |
2.2852 | 2500 | 0.0009 | - | 0.6557 |
2.3766 | 2600 | - | - | 0.6559 |
2.4680 | 2700 | - | - | 0.6560 |
2.5594 | 2800 | - | - | 0.6560 |
2.6508 | 2900 | - | - | 0.6564 |
2.7422 | 3000 | 0.0007 | - | 0.6565 |
2.8336 | 3100 | - | - | 0.6565 |
2.9250 | 3200 | - | - | 0.6564 |
3.0 | 3282 | - | - | 0.6565 |
3.0165 | 3300 | - | - | 0.6566 |
3.1079 | 3400 | - | - | 0.6568 |
3.1993 | 3500 | 0.0007 | - | 0.6565 |
3.2907 | 3600 | - | - | 0.6563 |
3.3821 | 3700 | - | - | 0.6564 |
3.4735 | 3800 | - | - | 0.6565 |
3.5649 | 3900 | - | - | 0.6565 |
3.6563 | 4000 | 0.0005 | - | 0.6566 |
3.7477 | 4100 | - | - | 0.6566 |
3.8391 | 4200 | - | - | 0.6566 |
3.9305 | 4300 | - | - | 0.6566 |
4.0 | 4376 | - | - | 0.6566 |
Framework Versions
- Python: 3.10.12
- Sentence Transformers: 3.0.1
- Transformers: 4.44.2
- PyTorch: 2.4.0+cu121
- Accelerate: 0.33.0
- Datasets: 2.21.0
- 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",
}
ContrastiveLoss
@inproceedings{hadsell2006dimensionality,
author={Hadsell, R. and Chopra, S. and LeCun, Y.},
booktitle={2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06)},
title={Dimensionality Reduction by Learning an Invariant Mapping},
year={2006},
volume={2},
number={},
pages={1735-1742},
doi={10.1109/CVPR.2006.100}
}