Hugging Face's logo

Northell
/
material-subdomain-classifier

Text Classification
setfit
Safetensors
sentence-transformers
bert
generated_from_setfit_trainer
Eval Results
Model card Files Community
material-subdomain-classifier / README.md
stan-northell's picture
stan-northell
Upload folder using huggingface_hub
5e32075 verified
preview code
|
raw
history blame
12.2 kB
metadata
tags:
  - setfit
  - sentence-transformers
  - text-classification
  - generated_from_setfit_trainer
widget:
  - text: >-
      hasAdditionalInformation: TFP10-73Pepsi Max X 264 TFP10-174 BAT Velo X
      353, hasColourDetails: 1pp - Face Print, hasCreatedDate: 2024-06-12,
      hasCustomerHomeCountry: United Kingdom, hasCustomerID: 25892,
      hasCustomerName: Co-operative Group Limited.(Co-operative Group Limited
      (Co-op Food)), hasCutting: Cut to shape, hasElementID: 3343462,
      hasElementTitle: POS028 SECURITY SHROUD, hasFinishedSizeHeight: 1540,
      hasFinishedSizeWidth: 600, hasFlatSizeHeight: 3080, hasFlatSizeWidth: 600,
      hasFscPaperBeenSpecified: No, hasInternalID:
      354d490a-f709-4034-af56-3e0b28ee34ba, hasMachineFinishing: Yes,
      hasMachineFinishingDetails: Trimmed to Size, Fold in Half, Weld Long Edges
      Only with 2 x PP Eyelets Positioned as Template - Fold Twice (to 600x515
      approx) for Flat Packing Pack in 2's, hasMaterialCategory: Plastic,
      hasMaterialDescription: 180gsm White/White Woven PE,
      hasMaterialThicknessOrWeight: 180, hasMaterialType: Polypropylene,
      hasMaterialUnitOfMeasure: GSM, hasNumberOfVersions: 2,
      hasPackingRequirements: Delivery to K Displays, Smith Way Ossett, FAO Dean
      Newbold. Delivery required Friday 21st June.  Please book in 48hrs in
      advance and mark all pallets on boxes with code, qty and P10 2024 Co-op
      Campaign, hasPrice: 3513.22 GBP, hasPrintedSides: Single sided,
      hasProofType: PDF digital proof, hasQuantity: 617,
      hasRecycledContentBeenOffered: No, hasSupplierName: Dominion Print
      Limited(Dominion Print Limited), hasTotalColours: 4, hasUnitOfMeasure:
      Millimetres (mm), 
  - text: >-
      hasAdditionalInformation: Mailed First Class, hasArtworkDoubleSidedStatus:
      Double Sided Different, hasCreatedDate: 2024-03-21,
      hasCustomerHomeCountry: United States, hasCustomerID: 32065,
      hasCustomerName: Republic Services, Inc(Republic Services), hasCutting:
      Trim to size, hasElementID: 3192439, hasElementTitle: Crockett Residental
      PC Mailer 2024, hasFinishedSizeHeight: 4, hasFinishedSizeWidth: 6,
      hasFscPaperBeenSpecified: No, hasInternalID:
      a63ca51f-99e2-4479-abb8-3e1f48c385e8, hasMaterialCategory: Paper,
      hasMaterialDescription: Uncoated Cover, hasMaterialThicknessOrWeight: 100,
      hasMaterialType: Paper, hasMaterialUnitOfMeasure: Pounds (lbs),
      hasNumberOfVersions: 1, hasPaperType: Cover, hasPrice: 302.6 USD,
      hasPrintedSides: Double sided, hasProofType: PDF digital proof,
      hasQuantity: 1200, hasRecycledContentBeenOffered: N/A, hasSendToDetails:
      [email protected], hasSupplierName: United Printing
      and Mail - HHG Strategic Partner (United Printing and Mail  - 48084 -
      HHGSP - US Only), hasTotalColours: 4, hasTotalColoursFace: 4,
      hasUnitOfMeasure: Inches (in), 
  - text: >-
      hasAdditionalInformation: US-89839_AIRSUPRA HCP Discover Leave behind Qt
      150,000 8.5”x11” flat/finished 80# Chorus Art Coated Cover 6/0 (CMYK +
      2PMS) + Satin AQ S/W in 25s, hasColourDetails: 6/0 (CMYK + 2PMS) + Satin
      AQ, hasCreatedDate: 2024-07-11, hasCustomerHomeCountry: United States,
      hasCustomerID: 31753, hasCustomerName: AstraZeneca Pharmaceuticals
      LP(AstraZeneca - US - BBU), hasCutting: Trim to size, hasElementID:
      3394425, hasElementTitle: US-89839_AIRSUPRA HCP Discover Leave behind,
      hasFinishedSizeHeight: 11, hasFinishedSizeWidth: 8.5, hasFlatSizeHeight:
      11, hasFlatSizeWidth: 8.5, hasFscPaperBeenSpecified: Yes, hasInternalID:
      91a64b08-cb2a-4d8e-b11d-b3908f11f2cd, hasMachineFinishing: Yes,
      hasMachineFinishingDetails: S/W in 25s, hasMaterialCategory: Paper,
      hasMaterialDescription: 80# Chorus Art Coated Cover,
      hasMaterialRecycledPercentage: 30%, hasMaterialThicknessOrWeight: 80,
      hasMaterialType: Paper and board, hasMaterialUnitOfMeasure: Pounds (lbs),
      hasNumberOfVersions: 1, hasPackingRequirements: S/W in 25s, hasPaperType:
      Cover, hasPrice: 13847.67 USD, hasPrintedSides: Single sided,
      hasProductCategory: Loose Print, hasProofType: PDF digital proof,Colour
      contract proof, hasQuantity: 150000, hasQuantityPerVersion: 150000,
      hasRecycledContentBeenOffered: Yes, hasSupplierName: Phoenix Lithographing
      Corporation(Phoenix Lithographing Corp - HHGSP - PI), hasTotalColours: 6,
      hasUnitOfMeasure: Inches (in), 
  - text: >-
      hasAdditionalInformation: US-82104_AIRSUPRA HCP Clinical Leave Behind Qt
      650,000 (4pg Bi-fold) 17"x11" flat 8.5"x11" finished 80# Coated Cover 6/6
      (CMYK + 2PMS) + GLOSS AQ Trim / Score / Bi-Fold S/W in 25s,
      hasArtworkDoubleSidedStatus: Double Sided Different, hasColourDetails: 6/6
      (CMYK + 2PMS) + GLOSS AQ, hasCreatedDate: 2024-01-18,
      hasCustomerHomeCountry: United States, hasCustomerID: 31753,
      hasCustomerName: AstraZeneca Pharmaceuticals LP(AstraZeneca - US - BBU),
      hasCutting: Trim to size, hasElementID: 3071417, hasElementTitle:
      US-82104_AIRSUPRA HCP Clinical Leave Behind, hasFinishedSizeHeight: 11,
      hasFinishedSizeWidth: 8.5, hasFlatSizeHeight: 11, hasFlatSizeWidth: 17,
      hasFscPaperBeenSpecified: Yes, hasInternalID:
      a8e77a84-d6af-4478-b83a-a54ea515b6f0, hasMachineFinishing: Yes,
      hasMachineFinishingDetails: Trim / Score / Bi-Fold S/W in 25s,
      hasMaterialCategory: Paper, hasMaterialDescription: 80# Coated Cover,
      hasMaterialRecycledPercentage: 0%, hasMaterialThicknessOrWeight: 80,
      hasMaterialType: Paper and board, hasMaterialUnitOfMeasure: Pounds (lbs),
      hasNumberOfVersions: 1, hasPackingRequirements: S/W in 25s, hasPaperType:
      Cover, hasPrice: 118754 USD, hasPrintedSides: Double sided,
      hasProductCategory: Booklets & Brochures, hasProofType: Colour contract
      proof,PDF digital proof, hasQuantity: 650000, hasQuantityPerVersion:
      650000, hasRecycledContentBeenOffered: Yes, hasSupplierName: Graphic Arts
      Incorporated(Graphic Arts Inc  - 56170 - HHGSP), hasTotalColours: 6,
      hasUnitOfMeasure: Inches (in), 
  - text: >-
      hasCreatedDate: 2024-01-04, hasCustomerHomeCountry: United States,
      hasCustomerID: 14458, hasCustomerName: Lowe's Companies Inc(Lowe's FVS),
      hasCutting: Trim to size, hasElementID: 3044623, hasElementTitle: G284515
      Commodity Moulding Profile Card 110911, hasFinishedSizeHeight: 6.875,
      hasFinishedSizeWidth: 3, hasFlatSizeHeight: 6.875, hasFlatSizeWidth: 3,
      hasFscPaperBeenSpecified: No, hasInternalID:
      c88f6dd9-5470-4870-a971-6d88eafb768d, hasMaterialCategory: Other,
      hasMaterialDescription: 8PT _C1S Cover, hasMaterialType: Other,
      hasNumberOfVersions: 1, hasPrice: 0.01 USD, hasPrintedSides: Single sided,
      hasProofType: PDF digital proof, hasQuantity: 1,
      hasRecycledContentBeenOffered: N/A, hasSupplierName: HH IC Content
      Production + Development(HH IC Content Production + Development),
      hasTotalColours: 4, hasUnitOfMeasure: Inches (in), 
metrics:
  - f1_micro
  - f1_macro
  - f1_weighted
  - precision
  - accuracy
  - recall
pipeline_tag: text-classification
library_name: setfit
inference: false
model-index:
  - name: SetFit
    results:
      - task:
          type: text-classification
          name: Text Classification
        dataset:
          name: Northell/ros-classifiers-materials-flat
          type: unknown
          split: test
        metrics:
          - type: f1_micro
            value: 0.4888472352389878
            name: F1_Micro
          - type: f1_macro
            value: 0.07490145637740193
            name: F1_Macro
          - type: f1_weighted
            value: 0.45529275569713784
            name: F1_Weighted
          - type: precision
            value: 0.8907103538513184
            name: Precision
          - type: accuracy
            value: 0.9836170077323914
            name: Accuracy
          - type: recall
            value: 0.33686384558677673
            name: Recall

SetFit

This is a SetFit model that can be used for Text Classification. A OneVsRestClassifier instance is used for classification.

The model has been trained using an efficient few-shot learning technique that involves:

  1. Fine-tuning a Sentence Transformer with contrastive learning.
  2. Training a classification head with features from the fine-tuned Sentence Transformer.

Model Details

Model Description

  • Model Type: SetFit
  • Classification head: a OneVsRestClassifier instance
  • Maximum Sequence Length: 512 tokens
  • Number of Classes: 43 classes

Model Sources

Evaluation

Metrics

Label F1_Micro F1_Macro F1_Weighted Precision Accuracy Recall
all 0.4888 0.0749 0.4553 0.8907 0.9836 0.3369

Uses

Direct Use for Inference

First install the SetFit library:

pip install setfit

Then you can load this model and run inference.

from setfit import SetFitModel

# Download from the 🤗 Hub
model = SetFitModel.from_pretrained("setfit_model_id")
# Run inference
preds = model("hasCreatedDate: 2024-01-04, hasCustomerHomeCountry: United States, hasCustomerID: 14458, hasCustomerName: Lowe's Companies Inc(Lowe's FVS), hasCutting: Trim to size, hasElementID: 3044623, hasElementTitle: G284515 Commodity Moulding Profile Card 110911, hasFinishedSizeHeight: 6.875, hasFinishedSizeWidth: 3, hasFlatSizeHeight: 6.875, hasFlatSizeWidth: 3, hasFscPaperBeenSpecified: No, hasInternalID: c88f6dd9-5470-4870-a971-6d88eafb768d, hasMaterialCategory: Other, hasMaterialDescription: 8PT _C1S Cover, hasMaterialType: Other, hasNumberOfVersions: 1, hasPrice: 0.01 USD, hasPrintedSides: Single sided, hasProofType: PDF digital proof, hasQuantity: 1, hasRecycledContentBeenOffered: N/A, hasSupplierName: HH IC Content Production + Development(HH IC Content Production + Development), hasTotalColours: 4, hasUnitOfMeasure: Inches (in), ")

Training Details

Training Set Metrics

Training set Min Median Max
Word count 61 109.9881 766

Framework Versions

  • Python: 3.10.16
  • SetFit: 1.1.1
  • Sentence Transformers: 3.4.1
  • Transformers: 4.49.0
  • PyTorch: 2.6.0+cu124
  • Datasets: 3.2.0
  • Tokenizers: 0.21.0

Citation

BibTeX 

@article{https://doi.org/10.48550/arxiv.2209.11055,
    doi = {10.48550/ARXIV.2209.11055},
    url = {https://arxiv.org/abs/2209.11055},
    author = {Tunstall, Lewis and Reimers, Nils and Jo, Unso Eun Seo and Bates, Luke and Korat, Daniel and Wasserblat, Moshe and Pereg, Oren},
    keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},
    title = {Efficient Few-Shot Learning Without Prompts},
    publisher = {arXiv},
    year = {2022},
    copyright = {Creative Commons Attribution 4.0 International}
}