langcache-embed-experimental / README.md

radoslavralev

Add new SentenceTransformer model

d50b42f verified about 1 month ago

preview code

raw

history blame

17.4 kB

metadata

language:
  - en
license: apache-2.0
tags:
  - biencoder
  - sentence-transformers
  - text-classification
  - sentence-pair-classification
  - semantic-similarity
  - semantic-search
  - retrieval
  - reranking
  - generated_from_trainer
  - dataset_size:13667
  - loss:ArcFaceInBatchLoss
base_model: sentence-transformers/all-MiniLM-L6-v2
widget:
  - source_sentence: >-
      It was mobilized in December 2014 from elements of the dissolved 51st
      Mechanized Brigade and newly formed units .
    sentences:
      - >-
        This North-South route falls entirely in the Belgian territory and runs
        together with the Belgian roads N31 and A17 .
      - >-
        It was mobilized in December 2014 from elements of the disbanded 51st
        Mechanized Brigade and newly formed units .
      - All windows are double wood , hung up with a single light .
  - source_sentence: It is located at Ellison Bay , in the town of Liberty Grove , Wisconsin .
    sentences:
      - >-
        It is located in Ellison Bay , in the town of Liberty Grove , Wisconsin
        .
      - >-
        It is located in Liberty Grove , Wisconsin , in the town of Ellison Bay
        .
      - >-
        The Hadejia River ( Hausa : `` kogin Haɗeja `` ) is a river in northern
        Nigeria and is a tributary of the Yobe River ( Komadugu Yobe ) .
  - source_sentence: >-
      Both long and short vowels can be nasalized ( differentiation between ``
      acces `` and `` Ä cces `` below ) , but long nasal vowels are more common
      .
    sentences:
      - >-
        Both long and short vowels can be nasalized ( the distinction between ``
        acces `` and `` ącces `` below ) , but long nasal vowels are more common
        .
      - >-
        Wilson was a member of the Senate from 1844 to 1846 and 1850 to 1852 .
        From 1851 to 1852 he was the Massachusetts State Senate 's President .
      - >-
        Both long vowels can be nasalized ( the distinction between `` acces ``
        and `` ącces `` below ) , but long and short nasal vowels are more
        common .
  - source_sentence: >-
      At that time , on June 22 , 1754 , Edward Bentham married Bentham
      Elizabeth Bates ( d . 1790 ) from Hampshire in the nearby county of Alton
      .
    sentences:
      - >-
        The Department of Criminal Justice developed the first certificate
        program in forensic science in North Carolina and sponsors a summer
        comparative studies program based in Europe .
      - >-
        At that time , on June 22 , 1754 , Edward Bentham married Bentham
        Elizabeth Bates ( d . 1790 ) from Hampshire in the nearby county of
        Alton .
      - >-
        It was at this time , on 22 June 1754 , that Edward Bentham married
        Elizabeth Bates ( d 1790 ) from Alton in the nearby county of Hampshire
        .
  - source_sentence: >-
      In 1973 Michels ' apos broke ; Barcelona the world transfer record to
      bring Cruyff to Catalonia .
    sentences:
      - >-
        In 1973 , Cruyff 'Barcelona broke the world transfer record to bring
        Michels to Catalonia .
      - >-
        Amalric then marched to Cairo , where Shawar offered Amalric two million
        pieces of gold .
      - >-
        In 1973 Michels ' apos broke ; Barcelona the world transfer record to
        bring Cruyff to Catalonia .
datasets:
  - redis/langcache-sentencepairs-v2
pipeline_tag: sentence-similarity
library_name: sentence-transformers
metrics:
  - cosine_accuracy@1
  - cosine_precision@1
  - cosine_recall@1
  - cosine_ndcg@10
  - cosine_mrr@1
  - cosine_map@100
  - cosine_auc_precision_cache_hit_ratio
  - cosine_auc_similarity_distribution
model-index:
  - name: Redis fine-tuned BiEncoder model for semantic caching on LangCache
    results:
      - task:
          type: custom-information-retrieval
          name: Custom Information Retrieval
        dataset:
          name: test
          type: test
        metrics:
          - type: cosine_accuracy@1
            value: 0.5767756724811061
            name: Cosine Accuracy@1
          - type: cosine_precision@1
            value: 0.5767756724811061
            name: Cosine Precision@1
          - type: cosine_recall@1
            value: 0.5587801563902068
            name: Cosine Recall@1
          - type: cosine_ndcg@10
            value: 0.765320607860921
            name: Cosine Ndcg@10
          - type: cosine_mrr@1
            value: 0.5767756724811061
            name: Cosine Mrr@1
          - type: cosine_map@100
            value: 0.7130569949974509
            name: Cosine Map@100
          - type: cosine_auc_precision_cache_hit_ratio
            value: 0.33372951540341317
            name: Cosine Auc Precision Cache Hit Ratio
          - type: cosine_auc_similarity_distribution
            value: 0.1529248551010913
            name: Cosine Auc Similarity Distribution

Redis fine-tuned BiEncoder model for semantic caching on LangCache

This is a sentence-transformers model finetuned from sentence-transformers/all-MiniLM-L6-v2 on the LangCache Sentence Pairs (all) dataset. It maps sentences & paragraphs to a 384-dimensional dense vector space and can be used for sentence pair similarity.

Model Details

Model Description

Model Type: Sentence Transformer
Base model: sentence-transformers/all-MiniLM-L6-v2
Maximum Sequence Length: 100 tokens
Output Dimensionality: 384 dimensions
Similarity Function: Cosine Similarity
Training Dataset:
- LangCache Sentence Pairs (all)
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': 100, 'do_lower_case': False, 'architecture': 'BertModel'})
  (1): Pooling({'word_embedding_dimension': 384, '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})
  (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("redis/langcache-embed-experimental")
# Run inference
sentences = [
    "In 1973 Michels ' apos broke ; Barcelona the world transfer record to bring Cruyff to Catalonia .",
    "In 1973 Michels ' apos broke ; Barcelona the world transfer record to bring Cruyff to Catalonia .",
    "In 1973 , Cruyff 'Barcelona broke the world transfer record to bring Michels to Catalonia .",
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 1.0000, 0.9219],
#         [1.0000, 1.0000, 0.9219],
#         [0.9219, 0.9219, 1.0078]], dtype=torch.bfloat16)

Evaluation

Metrics

Custom Information Retrieval

Dataset: test
Evaluated with ir_evaluator.CustomInformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.5768
cosine_precision@1	0.5768
cosine_recall@1	0.5588
cosine_ndcg@10	0.7653
cosine_mrr@1	0.5768
cosine_map@100	0.7131
cosine_auc_precision_cache_hit_ratio	0.3337
cosine_auc_similarity_distribution	0.1529

Training Details

Training Dataset

LangCache Sentence Pairs (all)

Dataset: LangCache Sentence Pairs (all)
Size: 6,780 training samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative
type	string	string	string
details	min: 8 tokens mean: 26.28 tokens max: 47 tokens	min: 8 tokens mean: 26.27 tokens max: 47 tokens	min: 8 tokens mean: 26.25 tokens max: 47 tokens

Samples:

anchor	positive	negative
`The newer Punts are still very much in existence today and race in the same fleets as the older boats .`	`The newer punts are still very much in existence today and run in the same fleets as the older boats .`	`This marine species occurs in the eastern Indian Ocean and before the Maldives and New Caledonia .`
`The newer punts are still very much in existence today and run in the same fleets as the older boats .`	`The newer Punts are still very much in existence today and race in the same fleets as the older boats .`	`Both young people burn with love really , for both , but without being able to say it to himself , admitting him always .`
`Turner Valley , was at the Turner Valley Bar N Ranch Airport , southwest of the Turner Valley Bar N Ranch , Alberta , Canada .`	`Turner Valley , , was located at Turner Valley Bar N Ranch Airport , southwest of Turner Valley Bar N Ranch , Alberta , Canada .`	`Turner Valley Bar N Ranch Airport , , was located at Turner Valley Bar N Ranch , southwest of Turner Valley , Alberta , Canada .`

Loss: losses.ArcFaceInBatchLoss with these parameters:

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

Evaluation Dataset

LangCache Sentence Pairs (all)

Dataset: LangCache Sentence Pairs (all)
Size: 6,780 evaluation samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative
type	string	string	string
details	min: 8 tokens mean: 26.28 tokens max: 47 tokens	min: 8 tokens mean: 26.27 tokens max: 47 tokens	min: 8 tokens mean: 26.25 tokens max: 47 tokens

Samples:

anchor	positive	negative
`The newer Punts are still very much in existence today and race in the same fleets as the older boats .`	`The newer punts are still very much in existence today and run in the same fleets as the older boats .`	`This marine species occurs in the eastern Indian Ocean and before the Maldives and New Caledonia .`
`The newer punts are still very much in existence today and run in the same fleets as the older boats .`	`The newer Punts are still very much in existence today and race in the same fleets as the older boats .`	`Both young people burn with love really , for both , but without being able to say it to himself , admitting him always .`
`Turner Valley , was at the Turner Valley Bar N Ranch Airport , southwest of the Turner Valley Bar N Ranch , Alberta , Canada .`	`Turner Valley , , was located at Turner Valley Bar N Ranch Airport , southwest of Turner Valley Bar N Ranch , Alberta , Canada .`	`Turner Valley Bar N Ranch Airport , , was located at Turner Valley Bar N Ranch , southwest of Turner Valley , Alberta , Canada .`

Loss: losses.ArcFaceInBatchLoss with these parameters:

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

Training Logs

Epoch	Step	test_cosine_ndcg@10
-1	-1	0.7653

Framework Versions

Python: 3.12.3
Sentence Transformers: 5.1.0
Transformers: 4.56.0
PyTorch: 2.8.0+cu128
Accelerate: 1.10.1
Datasets: 4.0.0
Tokenizers: 0.22.0

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",
}