Hugging Face's logo

faodl
/
model_cca_multilabel_MiniLM-L12-70prop-data-augmented

Text Classification
setfit
Safetensors
sentence-transformers
bert
generated_from_setfit_trainer
text-embeddings-inference
Model card Files
xet
Community
model_cca_multilabel_MiniLM-L12-70prop-data-augmented / README.md
lgsilvaesilva's picture
lgsilvaesilva
Push model using huggingface_hub.
69075f5 verified
preview code
|
raw
history blame contribute delete
24.2 kB
metadata 
tags:
  - setfit
  - sentence-transformers
  - text-classification
  - generated_from_setfit_trainer
widget:
  - text: >-
      Strengthen macro-fiscal resilience through risk-informed public investment
      planning, including scenario-based budgeting and contingent financing
      arrangements.
  - text: >-
      finding environmentally sustainable energy solutions is central to the
      document. it seeks to facilitate cultural, institutional and technological
      change in a way that supports ''aggressive'' advances in energy efficiency
      and conservation, minimises greenhouse emissions and ultimately provides
      green growth. these energy efficiency and conservation goals are seen as
      ''no regrets'' mitigation actions that can have positive impacts on
      society and the economy, principally by reducing costs and dependency on
      fossil fuel imports. overall the policy propose to reduce the percentage
      of petroleum in the country''s energy supply mix from the current 95
      percent (does not state to what level) and increase the percentage of
      renewables in the energy mix with proposed targets of 11 percent by 2012,
      12.5 percent by 2015 and 20 percent by 2030. six sub-policies exist to
      support the national energy policy, namely: - a carbon emissions trading
      policy developed to address jamaica''s participation in the clean
      development mechanism - energy-from-waste policy - national renewable
      energy policy 2010-2030 - national energy from waste policy 2010-2030 -
      energy conservation and efficiency policy - biofuels policy
  - text: >-
      objetivos: 1. promover la garantía del derecho a la alimentación para la
      población general y en especial para las personas y grupos de mayor
      vulnerabilidad. 2. respetar la identidad cultural, las necesidades
      nutricionales según el ciclo de vida y la diversidad de formas de
      producción, de consumo y comercialización agropecuaria, fortaleciendo los
      mercados locales, sin contraponerse al comercio agroalimentario
      internacional, favoreciéndose la producción nacional en granos básicos,
      frutas y vegetales. 3. promover la igualdad entre hombres y mujeres, dando
      las mismas posibilidades de acceso a recursos productivos, servicios y
      oportunidades para asumir responsabilidades y roles en la seguridad
      alimentaria y nutricional. 4.transformar el enfoque de las políticas
      públicas y sociales, para que pasen las personas de ser clientela pasiva y
      vulnerable que requiere de asistencia, a personas sujetos de derechos.
  - text: >-
      Regulatory arrangements will be reformed to accelerate innovation in
      agriculture, including pilot programs, regulatory sandboxes for new inputs
      and services, clear intellectual property protection, and predictable
      approval timelines for agrochemical and digital solutions that meet safety
      and environmental criteria.
  - text: >-
      Climate-smart strategies will protect livelihoods by diversifying income
      sources, expanding agroforestry and drought-resistant crops, and
      implementing risk-transfer mechanisms that shield poor households from
      shocks, thereby contributing to sustained declines in poverty levels.
metrics:
  - accuracy
pipeline_tag: text-classification
library_name: setfit
inference: false
base_model: sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2

SetFit with sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2

This is a SetFit model that can be used for Text Classification. This SetFit model uses sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2 as the Sentence Transformer embedding model. 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 Sources

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("faodl/model_cca_multilabel_MiniLM-L12-70prop-data-augmented")
# Run inference
preds = model("Strengthen macro-fiscal resilience through risk-informed public investment planning, including scenario-based budgeting and contingent financing arrangements.")

Training Details

Training Set Metrics

Training set Min Median Max
Word count 1 69.0403 951

Training Hyperparameters

  • batch_size: (16, 16)
  • num_epochs: (2, 2)
  • max_steps: -1
  • sampling_strategy: oversampling
  • num_iterations: 20
  • body_learning_rate: (2e-05, 2e-05)
  • head_learning_rate: 2e-05
  • loss: CosineSimilarityLoss
  • distance_metric: cosine_distance
  • margin: 0.25
  • end_to_end: False
  • use_amp: False
  • warmup_proportion: 0.1
  • l2_weight: 0.01
  • seed: 42
  • eval_max_steps: -1
  • load_best_model_at_end: False

Training Results

Epoch Step Training Loss Validation Loss
0.0001 1 0.2247 -
0.0065 50 0.2105 -
0.0130 100 0.1984 -
0.0195 150 0.1899 -
0.0260 200 0.1916 -
0.0325 250 0.1769 -
0.0390 300 0.1679 -
0.0455 350 0.1677 -
0.0520 400 0.1591 -
0.0585 450 0.1521 -
0.0650 500 0.1522 -
0.0715 550 0.1497 -
0.0780 600 0.1494 -
0.0845 650 0.1457 -
0.0910 700 0.1503 -
0.0975 750 0.1328 -
0.1040 800 0.1251 -
0.1105 850 0.1395 -
0.1170 900 0.1298 -
0.1235 950 0.1221 -
0.1300 1000 0.1313 -
0.1365 1050 0.1267 -
0.1429 1100 0.1367 -
0.1494 1150 0.1324 -
0.1559 1200 0.1201 -
0.1624 1250 0.1244 -
0.1689 1300 0.1231 -
0.1754 1350 0.1214 -
0.1819 1400 0.1098 -
0.1884 1450 0.1152 -
0.1949 1500 0.1149 -
0.2014 1550 0.1185 -
0.2079 1600 0.1123 -
0.2144 1650 0.1092 -
0.2209 1700 0.1097 -
0.2274 1750 0.1159 -
0.2339 1800 0.1076 -
0.2404 1850 0.114 -
0.2469 1900 0.1055 -
0.2534 1950 0.1033 -
0.2599 2000 0.1016 -
0.2664 2050 0.1004 -
0.2729 2100 0.0973 -
0.2794 2150 0.1051 -
0.2859 2200 0.0954 -
0.2924 2250 0.0998 -
0.2989 2300 0.0984 -
0.3054 2350 0.0906 -
0.3119 2400 0.0939 -
0.3184 2450 0.1023 -
0.3249 2500 0.0983 -
0.3314 2550 0.0952 -
0.3379 2600 0.099 -
0.3444 2650 0.0994 -
0.3509 2700 0.0975 -
0.3574 2750 0.0871 -
0.3639 2800 0.0969 -
0.3704 2850 0.0845 -
0.3769 2900 0.1007 -
0.3834 2950 0.0887 -
0.3899 3000 0.0807 -
0.3964 3050 0.0859 -
0.4029 3100 0.0826 -
0.4094 3150 0.0784 -
0.4159 3200 0.0851 -
0.4224 3250 0.0834 -
0.4288 3300 0.0922 -
0.4353 3350 0.0862 -
0.4418 3400 0.0856 -
0.4483 3450 0.0848 -
0.4548 3500 0.0735 -
0.4613 3550 0.0752 -
0.4678 3600 0.0881 -
0.4743 3650 0.0836 -
0.4808 3700 0.0808 -
0.4873 3750 0.0963 -
0.4938 3800 0.0816 -
0.5003 3850 0.0809 -
0.5068 3900 0.0833 -
0.5133 3950 0.0852 -
0.5198 4000 0.0788 -
0.5263 4050 0.0742 -
0.5328 4100 0.0693 -
0.5393 4150 0.0856 -
0.5458 4200 0.072 -
0.5523 4250 0.0805 -
0.5588 4300 0.0741 -
0.5653 4350 0.0845 -
0.5718 4400 0.0753 -
0.5783 4450 0.0814 -
0.5848 4500 0.0691 -
0.5913 4550 0.0823 -
0.5978 4600 0.0847 -
0.6043 4650 0.0714 -
0.6108 4700 0.0879 -
0.6173 4750 0.0711 -
0.6238 4800 0.0697 -
0.6303 4850 0.0741 -
0.6368 4900 0.0771 -
0.6433 4950 0.0837 -
0.6498 5000 0.0743 -
0.6563 5050 0.0755 -
0.6628 5100 0.0739 -
0.6693 5150 0.0816 -
0.6758 5200 0.0782 -
0.6823 5250 0.0755 -
0.6888 5300 0.0712 -
0.6953 5350 0.0639 -
0.7018 5400 0.0694 -
0.7083 5450 0.0806 -
0.7147 5500 0.071 -
0.7212 5550 0.0707 -
0.7277 5600 0.0751 -
0.7342 5650 0.0724 -
0.7407 5700 0.0688 -
0.7472 5750 0.067 -
0.7537 5800 0.0718 -
0.7602 5850 0.0681 -
0.7667 5900 0.0694 -
0.7732 5950 0.0693 -
0.7797 6000 0.0731 -
0.7862 6050 0.0626 -
0.7927 6100 0.0691 -
0.7992 6150 0.0711 -
0.8057 6200 0.0627 -
0.8122 6250 0.0726 -
0.8187 6300 0.068 -
0.8252 6350 0.0766 -
0.8317 6400 0.0617 -
0.8382 6450 0.0671 -
0.8447 6500 0.0645 -
0.8512 6550 0.0722 -
0.8577 6600 0.0751 -
0.8642 6650 0.0591 -
0.8707 6700 0.0664 -
0.8772 6750 0.0735 -
0.8837 6800 0.0709 -
0.8902 6850 0.0632 -
0.8967 6900 0.0679 -
0.9032 6950 0.0596 -
0.9097 7000 0.0676 -
0.9162 7050 0.066 -
0.9227 7100 0.069 -
0.9292 7150 0.0615 -
0.9357 7200 0.0579 -
0.9422 7250 0.0576 -
0.9487 7300 0.0558 -
0.9552 7350 0.0556 -
0.9617 7400 0.0637 -
0.9682 7450 0.0615 -
0.9747 7500 0.0677 -
0.9812 7550 0.0584 -
0.9877 7600 0.0661 -
0.9942 7650 0.0583 -
1.0006 7700 0.0639 -
1.0071 7750 0.0598 -
1.0136 7800 0.0586 -
1.0201 7850 0.055 -
1.0266 7900 0.0636 -
1.0331 7950 0.0623 -
1.0396 8000 0.0661 -
1.0461 8050 0.0633 -
1.0526 8100 0.056 -
1.0591 8150 0.0555 -
1.0656 8200 0.0608 -
1.0721 8250 0.0491 -
1.0786 8300 0.0592 -
1.0851 8350 0.0645 -
1.0916 8400 0.0553 -
1.0981 8450 0.0547 -
1.1046 8500 0.0494 -
1.1111 8550 0.0594 -
1.1176 8600 0.058 -
1.1241 8650 0.0589 -
1.1306 8700 0.0552 -
1.1371 8750 0.0554 -
1.1436 8800 0.0566 -
1.1501 8850 0.0558 -
1.1566 8900 0.0596 -
1.1631 8950 0.0551 -
1.1696 9000 0.061 -
1.1761 9050 0.0689 -
1.1826 9100 0.0565 -
1.1891 9150 0.0581 -
1.1956 9200 0.0606 -
1.2021 9250 0.057 -
1.2086 9300 0.0577 -
1.2151 9350 0.0629 -
1.2216 9400 0.0592 -
1.2281 9450 0.0547 -
1.2346 9500 0.0606 -
1.2411 9550 0.0588 -
1.2476 9600 0.0581 -
1.2541 9650 0.0624 -
1.2606 9700 0.0589 -
1.2671 9750 0.0646 -
1.2736 9800 0.0559 -
1.2801 9850 0.0594 -
1.2865 9900 0.0586 -
1.2930 9950 0.0552 -
1.2995 10000 0.0513 -
1.3060 10050 0.0565 -
1.3125 10100 0.0626 -
1.3190 10150 0.0483 -
1.3255 10200 0.0643 -
1.3320 10250 0.0524 -
1.3385 10300 0.0559 -
1.3450 10350 0.0589 -
1.3515 10400 0.0562 -
1.3580 10450 0.0592 -
1.3645 10500 0.047 -
1.3710 10550 0.0531 -
1.3775 10600 0.0506 -
1.3840 10650 0.0579 -
1.3905 10700 0.0569 -
1.3970 10750 0.0579 -
1.4035 10800 0.0504 -
1.4100 10850 0.0547 -
1.4165 10900 0.0497 -
1.4230 10950 0.0533 -
1.4295 11000 0.0488 -
1.4360 11050 0.0537 -
1.4425 11100 0.0544 -
1.4490 11150 0.0548 -
1.4555 11200 0.0475 -
1.4620 11250 0.0519 -
1.4685 11300 0.0568 -
1.4750 11350 0.0567 -
1.4815 11400 0.0473 -
1.4880 11450 0.0535 -
1.4945 11500 0.0531 -
1.5010 11550 0.0567 -
1.5075 11600 0.0529 -
1.5140 11650 0.0544 -
1.5205 11700 0.0612 -
1.5270 11750 0.055 -
1.5335 11800 0.0474 -
1.5400 11850 0.0572 -
1.5465 11900 0.0484 -
1.5530 11950 0.0553 -
1.5595 12000 0.0519 -
1.5660 12050 0.0565 -
1.5724 12100 0.0466 -
1.5789 12150 0.0502 -
1.5854 12200 0.0525 -
1.5919 12250 0.054 -
1.5984 12300 0.0556 -
1.6049 12350 0.0515 -
1.6114 12400 0.0476 -
1.6179 12450 0.0579 -
1.6244 12500 0.0567 -
1.6309 12550 0.0551 -
1.6374 12600 0.0518 -
1.6439 12650 0.0508 -
1.6504 12700 0.0503 -
1.6569 12750 0.0484 -
1.6634 12800 0.0531 -
1.6699 12850 0.0553 -
1.6764 12900 0.0588 -
1.6829 12950 0.0547 -
1.6894 13000 0.0587 -
1.6959 13050 0.0562 -
1.7024 13100 0.0558 -
1.7089 13150 0.0559 -
1.7154 13200 0.0547 -
1.7219 13250 0.059 -
1.7284 13300 0.053 -
1.7349 13350 0.0532 -
1.7414 13400 0.0552 -
1.7479 13450 0.0443 -
1.7544 13500 0.058 -
1.7609 13550 0.0503 -
1.7674 13600 0.0499 -
1.7739 13650 0.0478 -
1.7804 13700 0.0569 -
1.7869 13750 0.052 -
1.7934 13800 0.0458 -
1.7999 13850 0.0551 -
1.8064 13900 0.0567 -
1.8129 13950 0.0511 -
1.8194 14000 0.0546 -
1.8259 14050 0.058 -
1.8324 14100 0.0539 -
1.8389 14150 0.0544 -
1.8454 14200 0.061 -
1.8519 14250 0.0521 -
1.8583 14300 0.046 -
1.8648 14350 0.0494 -
1.8713 14400 0.0604 -
1.8778 14450 0.0543 -
1.8843 14500 0.0522 -
1.8908 14550 0.0533 -
1.8973 14600 0.0469 -
1.9038 14650 0.0525 -
1.9103 14700 0.0516 -
1.9168 14750 0.0485 -
1.9233 14800 0.0601 -
1.9298 14850 0.0487 -
1.9363 14900 0.0496 -
1.9428 14950 0.0529 -
1.9493 15000 0.054 -
1.9558 15050 0.0431 -
1.9623 15100 0.0449 -
1.9688 15150 0.0602 -
1.9753 15200 0.0447 -
1.9818 15250 0.0506 -
1.9883 15300 0.0503 -
1.9948 15350 0.0515 -

Framework Versions

  • Python: 3.12.12
  • SetFit: 1.1.3
  • Sentence Transformers: 5.1.1
  • Transformers: 4.57.1
  • PyTorch: 2.8.0+cu126
  • Datasets: 4.0.0
  • Tokenizers: 0.22.1

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}
}