metadata

tags:
  - setfit
  - sentence-transformers
  - text-classification
  - generated_from_setfit_trainer
widget:
  - text: >-
      WHO and UNICEF has recommended that a child should receive the minimum
      dietary diversity (MDD) of foods and beverages from at least five out of
      eight defined food groups to maintain proper growth and development during
      this critical period 19 . In Timor-Leste, 35.3% received minimum dietary
      diversity (MDD) 4 . On the other hand, the proportion of children 6-23
      months receiving MDD has been on the upward rise (28% in 2013 to 35.3% in
      2020) although it is still low. Food group diversity is associated with
      improved linear growth in young children20 . A diet lacking in diversity
      can increase the risk of micronutrient deficiencies, which may have a
      damaging effect on 47.0% 81.7% 93.4% 75.2% 30.7% 57.5% 62.3% 50.2% 0% 10%
      20% 30% 40% 50% 60% 70% 80% 90% 100% TLDHS 2003 TLDHS 2010 TLFNS 2013
      TLFNS 2016 46.8% 64.2% TLFNS 2020 Early Initiation (1 hour) Exclusive
      breastfeeding (0-5 months) 20NATIONAL HEALTH SECTOR NUTRITION STRATEGIC
      PLAN 2022-2026 children’s physical and cognitive development21 .
      Consequently, TLFNS 2020 reported that a very high proportion of children
      6-23 months had consumed grains, roots, and tubers (97.5%) and breast milk
      (90.6%), as well as vitamin A-rich fruits and vegetables (71.5%).
      Consumption of dairy products (0.8%) was low, while consumption of flesh
      foods (23.1%) and legumes or nuts (31.0%) was also relatively low. The
      2020 survey reported that 19.1% of children 6-23 months consumed sugar
      sweetened beverages, 31.0% consumed sweet or savoury junk foods, while
      20.0% did not consume any fruits or vegetables and 35.9% consumed no eggs
      or flesh foods.
  - text: >-
      Climate Risk and Vulnerability Baseline. One of the key roles of the NAP
      process is to develop a common evidence base on CC that can be referenced
      by stakeholders in various documents, including strategies and project
      proposals. Therefore, climate risk and vulnerability assessments shall be
      summarized and updated on a periodical basis to underlie the development
      of the NAP and the list of m
  - text: >-
      Agriculture in Armenia has always been remarkable with the high level of
      climate risks (hail damage, frost damage, drought, etc.). As it is already
      mentioned, agriculture has suffered losses from natural disasters worth of
      AMD 110 billion during the recent 6 years. Climate risks in Armenia are a
      serious problem since there are no clearly formed such state, political or
      institutional mechanisms, the application of which would make it possible
      to noticeably mitigate the existing risks. Due to the lack of such
      mechanisms, the mechanism of full assessment of the agricultural losses
      does not work too, as well as the risks are not assessed in advance. In
      this context, to reduce the agricultural risks, to introduce loss
      compensation mechanisms in a systemized way, and to provide sustainable
      income levels for economic entities, it is necessary to address the
      critical issue of agricultural risk insurance. 
  - text: >-
      Strategy 6.3: Strengthen monitoring, evaluation and surveillance systems
      for routine information sharing and data utilization at all levels
      Activities Stakeholder Conduct bi-annual nutrition M&E coordination
      meetings. ND, M&ED, INS Collaborate with HIS Department (HISD) and M&E
      Department MOH to conduct routine nutrition data quality assessments and
      audits (RDQA). ND, HISD, M&ED, INS In collaboration with HISD MOH and M&E
      Department, train M&E officers, DPHO nutrition, nutrition focal points and
      Municipality Health Services on data management (collection analyses,
      interpreting and reporting) at all levels. ND, HISD, M&ED, INS Develop and
      disseminate the Nutrition M&E Plan. ND, M&ED Strengthen the nutrition
      information system within the HMIS by integrating key nutrition indicators
      and databases. ND, HISD, M&ED Establish and scale up a nutrition
      surveillance system for real time monitoring at all levels. ND, M&ED, INS
      Conduct mid-term and end-term evaluation of the nutrition strategic plan.
      ND, HISD, M&ED, INS Conduct a food and nutrition survey every 5 years. ND,
      HISD, M&ED, INS Conduct knowledge attitude and practices (KAP) survey on
      nutrition. ND, HISD, M&ED, HPD, INS Liaise with HMIS to introduce
      real-time data collection linked to DHIS2. ND, HISD, M&ED Periodic
      publishing of nutrition bulletin/report ND, HISD, M&ED Develop and
      regularly review nutrition indicators monitoring and evaluation guideline.
      ND, HMIS, M&ED, INS 
  - text: >-
      Provision 1 - Access to safe nutritious food for all The package will be
      aimed at ending hunger and all forms of malnutrition and reduce the
      incidence of non-communicable diseases, enabling all people to be
      nourished and healthy. This suggests that all people at all times have
      access to sufficient quantities of affordable and safe foo
metrics:
  - accuracy
pipeline_tag: text-classification
library_name: setfit
inference: false
base_model: sentence-transformers/paraphrase-mpnet-base-v2

SetFit with sentence-transformers/paraphrase-mpnet-base-v2

This is a SetFit model that can be used for Text Classification. This SetFit model uses sentence-transformers/paraphrase-mpnet-base-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:

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

Model Details

Model Description

Model Type: SetFit
Sentence Transformer body: sentence-transformers/paraphrase-mpnet-base-v2
Classification head: a OneVsRestClassifier instance
Maximum Sequence Length: 512 tokens

Model Sources

Repository: SetFit on GitHub
Paper: Efficient Few-Shot Learning Without Prompts
Blogpost: SetFit: Efficient Few-Shot Learning Without Prompts

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/setfit-paraphrase-mpnet-base-v2-5ClassesDesc-multilabel")
# Run inference
preds = model("Provision 1 - Access to safe nutritious food for all The package will be aimed at ending hunger and all forms of malnutrition and reduce the incidence of non-communicable diseases, enabling all people to be nourished and healthy. This suggests that all people at all times have access to sufficient quantities of affordable and safe foo")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	7	123.3475	1014

Training Hyperparameters

batch_size: (8, 8)
num_epochs: (1, 1)
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
0.0028	1	0.3314	-
0.0709	50	0.2212	-
0.1418	100	0.1679	-
0.2128	150	0.1224	-
0.2837	200	0.0782	-
0.3546	250	0.0889	-
0.4255	300	0.0765	-
0.4965	350	0.0591	-
0.5674	400	0.0511	-
0.6383	450	0.0364	-
0.7092	500	0.0454	-
0.7801	550	0.0327	-
0.8511	600	0.0237	-
0.9220	650	0.024	-
0.9929	700	0.0216	-

Framework Versions

Python: 3.11.11
SetFit: 1.1.1
Sentence Transformers: 3.4.1
Transformers: 4.50.2
PyTorch: 2.6.0+cu124
Datasets: 3.5.0
Tokenizers: 0.21.1

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}
}