metadata

library_name: setfit
tags:
  - setfit
  - sentence-transformers
  - text-classification
  - generated_from_setfit_trainer
metrics:
  - accuracy
widget:
  - text: >-
      first: We recommend employees start a support group to share and address
      workplace concerns. second: Grievance Resolution Committee: A committee
      addresses formal grievances and ensures a fair resolution process.
  - text: >-
      first: Supervisors are encouraged to watch TED talks on communication to
      enhance their skills. second: Progressive Discipline: Disciplinary actions
      are proportionate and follow a structured process.
  - text: >-
      first: Grievance Resolution Committee: A committee addresses formal
      grievances and ensures a fair resolution process. second: We provide
      employees with a comprehensive handbook outlining our dispute resolution
      process for clarity.
  - text: >-
      first: We recommend employees seek advice from their peers and mentors to
      navigate workplace issues. second: We use technology-based solutions to
      facilitate virtual conflict resolution discussions.
  - text: >-
      first: We've introduced a complaint of the month contest to highlight and
      address concerns effectively. second: Our company has a clear conflict
      resolution policy that all employees must follow.
pipeline_tag: text-classification
inference: true
base_model: sentence-transformers/all-MiniLM-L6-v2
model-index:
  - name: SetFit with sentence-transformers/all-MiniLM-L6-v2
    results:
      - task:
          type: text-classification
          name: Text Classification
        dataset:
          name: Unknown
          type: unknown
          split: test
        metrics:
          - type: accuracy
            value: 0.7272727272727273
            name: Accuracy

SetFit with sentence-transformers/all-MiniLM-L6-v2

This is a SetFit model that can be used for Text Classification. This SetFit model uses sentence-transformers/all-MiniLM-L6-v2 as the Sentence Transformer embedding model. A LogisticRegression 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/all-MiniLM-L6-v2
Classification head: a LogisticRegression instance
Maximum Sequence Length: 256 tokens
Number of Classes: 2 classes

Model Sources

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

Model Labels

Label	Examples
0	'first: Employee Support Groups: Peer-led support groups for employees facing similar issues. second: We offer conflict resolution workshops to provide employees with valuable skills.'
1	'first: Conflict Resolution Peer Mentoring: Experienced employees mentor newcomers in conflict resolution. second: Diversity and Inclusion Training: Programs that promote understanding and reduce conflicts related to diversity.'

Evaluation

Metrics

Label	Accuracy
all	0.7273

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("sijan1/setfit-finetuned-fairness")
# Run inference
preds = model("first: We've introduced a complaint of the month contest to highlight and address concerns effectively. second: Our company has a clear conflict resolution policy that all employees must follow.")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	24	25.5	27

Label	Training Sample Count
0	1
1	1

Training Hyperparameters

batch_size: (4, 4)
num_epochs: (1, 1)
max_steps: -1
sampling_strategy: oversampling
num_iterations: 30
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
seed: 42
eval_max_steps: -1
load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.1	1	0.0141	-
5.0	50	0.0012	-
10.0	100	0.0006	-
0.1	1	0.0005	-
5.0	50	0.0005	-
10.0	100	0.0002	-
15.0	150	0.0002	-
20.0	200	0.0001	-
25.0	250	0.0001	-
30.0	300	0.0001	-
35.0	350	0.0002	-
40.0	400	0.0	-
45.0	450	0.0	-
50.0	500	0.0	-
55.0	550	0.0	-
60.0	600	0.0	-
65.0	650	0.0001	-
70.0	700	0.0	-
75.0	750	0.0	-
80.0	800	0.0	-
85.0	850	0.0	-
90.0	900	0.0	-
95.0	950	0.0	-
100.0	1000	0.0	-
0.0667	1	0.0	-
0.8333	50	0.0	-
0.0222	1	0.0	-
0.8333	50	0.0	-
0.0111	1	0.0001	-
0.5556	50	0.0	-
0.0333	1	0.0	-

Framework Versions

Python: 3.10.12
SetFit: 1.0.3
Sentence Transformers: 2.3.1
Transformers: 4.37.2
PyTorch: 2.1.0+cu121
Datasets: 2.17.1
Tokenizers: 0.15.2

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}
}