metadata
library_name: setfit
tags:
- setfit
- sentence-transformers
- text-classification
- generated_from_setfit_trainer
metrics:
- accuracy
widget:
- text: aku hanya menyukai setiap menit film ini.
- text: bioskop orang dalam kondisi terbaiknya.
- text: >-
bukan untuk orang yang mudah tersinggung atau mudah tersinggung, ini
adalah pemeriksaan yang berani dan berkepanjangan terhadap budaya yang
diidolakan, kebencian terhadap diri sendiri, dan politik seksual.
- text: itu curang.
- text: >-
Meskipun penduduk setempat akan senang melihat situs-situs Cleveland,
seluruh dunia akan menikmati komedi bertempo cepat dengan keunikan yang
mungkin membuat iri para coen bersaudara yang telah memenangkan
penghargaan.
pipeline_tag: text-classification
inference: true
base_model: firqaaa/indo-sentence-bert-base
model-index:
- name: SetFit with firqaaa/indo-sentence-bert-base
results:
- task:
type: text-classification
name: Text Classification
dataset:
name: Unknown
type: unknown
split: test
metrics:
- type: accuracy
value: 0.3425339366515837
name: Accuracy
SetFit with firqaaa/indo-sentence-bert-base
This is a SetFit model that can be used for Text Classification. This SetFit model uses firqaaa/indo-sentence-bert-base as the Sentence Transformer embedding model. A SetFitHead 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: firqaaa/indo-sentence-bert-base
- Classification head: a SetFitHead instance
- Maximum Sequence Length: 512 tokens
- Number of Classes: 5 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 |
---|---|
sangat positif |
|
sangat negatif |
|
positif |
|
negatif |
|
netral |
|
Evaluation
Metrics
Label | Accuracy |
---|---|
all | 0.3425 |
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("firqaaa/indo-setfit-bert-base-p2")
# Run inference
preds = model("itu curang.")
Training Details
Training Set Metrics
Training set | Min | Median | Max |
---|---|---|---|
Word count | 2 | 15.476 | 46 |
Label | Training Sample Count |
---|---|
sangat negatif | 200 |
negatif | 200 |
netral | 200 |
positif | 200 |
sangat positif | 200 |
Training Hyperparameters
- batch_size: (128, 32)
- num_epochs: (1, 8)
- max_steps: -1
- sampling_strategy: oversampling
- body_learning_rate: (2e-05, 5e-06)
- 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: True
Training Results
Epoch | Step | Training Loss | Validation Loss |
---|---|---|---|
0.0002 | 1 | 0.3317 | - |
0.008 | 50 | 0.2883 | - |
0.016 | 100 | 0.2625 | - |
0.024 | 150 | 0.2516 | - |
0.032 | 200 | 0.2075 | - |
0.04 | 250 | 0.184 | - |
0.048 | 300 | 0.1632 | - |
0.056 | 350 | 0.1105 | - |
0.064 | 400 | 0.1109 | - |
0.072 | 450 | 0.0934 | - |
0.08 | 500 | 0.0518 | - |
0.088 | 550 | 0.0246 | - |
0.096 | 600 | 0.0133 | - |
0.104 | 650 | 0.0056 | - |
0.112 | 700 | 0.006 | - |
0.12 | 750 | 0.0072 | - |
0.128 | 800 | 0.0179 | - |
0.136 | 850 | 0.0025 | - |
0.144 | 900 | 0.0019 | - |
0.152 | 950 | 0.0008 | - |
0.16 | 1000 | 0.0009 | - |
0.168 | 1050 | 0.0016 | - |
0.176 | 1100 | 0.0008 | - |
0.184 | 1150 | 0.0009 | - |
0.192 | 1200 | 0.0006 | - |
0.2 | 1250 | 0.0112 | - |
0.208 | 1300 | 0.0007 | - |
0.216 | 1350 | 0.0005 | - |
0.224 | 1400 | 0.0006 | - |
0.232 | 1450 | 0.0004 | - |
0.24 | 1500 | 0.0003 | - |
0.248 | 1550 | 0.0111 | - |
0.256 | 1600 | 0.0007 | - |
0.264 | 1650 | 0.0004 | - |
0.272 | 1700 | 0.0068 | - |
0.28 | 1750 | 0.0006 | - |
0.288 | 1800 | 0.008 | - |
0.296 | 1850 | 0.0004 | - |
0.304 | 1900 | 0.0009 | - |
0.312 | 1950 | 0.0004 | - |
0.32 | 2000 | 0.0003 | - |
0.328 | 2050 | 0.0034 | - |
0.336 | 2100 | 0.0003 | - |
0.344 | 2150 | 0.0002 | - |
0.352 | 2200 | 0.0002 | - |
0.36 | 2250 | 0.0002 | - |
0.368 | 2300 | 0.0002 | - |
0.376 | 2350 | 0.0002 | - |
0.384 | 2400 | 0.0002 | - |
0.392 | 2450 | 0.0001 | - |
0.4 | 2500 | 0.0002 | - |
0.408 | 2550 | 0.0001 | - |
0.416 | 2600 | 0.0001 | - |
0.424 | 2650 | 0.0002 | - |
0.432 | 2700 | 0.0001 | - |
0.44 | 2750 | 0.0001 | - |
0.448 | 2800 | 0.0001 | - |
0.456 | 2850 | 0.0003 | - |
0.464 | 2900 | 0.0001 | - |
0.472 | 2950 | 0.0001 | - |
0.48 | 3000 | 0.0004 | - |
0.488 | 3050 | 0.0002 | - |
0.496 | 3100 | 0.0001 | - |
0.504 | 3150 | 0.0003 | - |
0.512 | 3200 | 0.0001 | - |
0.52 | 3250 | 0.0001 | - |
0.528 | 3300 | 0.0002 | - |
0.536 | 3350 | 0.0001 | - |
0.544 | 3400 | 0.0001 | - |
0.552 | 3450 | 0.0001 | - |
0.56 | 3500 | 0.0001 | - |
0.568 | 3550 | 0.0001 | - |
0.576 | 3600 | 0.0001 | - |
0.584 | 3650 | 0.0001 | - |
0.592 | 3700 | 0.0001 | - |
0.6 | 3750 | 0.0 | - |
0.608 | 3800 | 0.0001 | - |
0.616 | 3850 | 0.0001 | - |
0.624 | 3900 | 0.0001 | - |
0.632 | 3950 | 0.0001 | - |
0.64 | 4000 | 0.0003 | - |
0.648 | 4050 | 0.0001 | - |
0.656 | 4100 | 0.0001 | - |
0.664 | 4150 | 0.0001 | - |
0.672 | 4200 | 0.0001 | - |
0.68 | 4250 | 0.0001 | - |
0.688 | 4300 | 0.0001 | - |
0.696 | 4350 | 0.0001 | - |
0.704 | 4400 | 0.0001 | - |
0.712 | 4450 | 0.0001 | - |
0.72 | 4500 | 0.0001 | - |
0.728 | 4550 | 0.0001 | - |
0.736 | 4600 | 0.0001 | - |
0.744 | 4650 | 0.0001 | - |
0.752 | 4700 | 0.0001 | - |
0.76 | 4750 | 0.0001 | - |
0.768 | 4800 | 0.0001 | - |
0.776 | 4850 | 0.0001 | - |
0.784 | 4900 | 0.0001 | - |
0.792 | 4950 | 0.0001 | - |
0.8 | 5000 | 0.0 | - |
0.808 | 5050 | 0.0001 | - |
0.816 | 5100 | 0.0001 | - |
0.824 | 5150 | 0.0001 | - |
0.832 | 5200 | 0.0 | - |
0.84 | 5250 | 0.0001 | - |
0.848 | 5300 | 0.0001 | - |
0.856 | 5350 | 0.0 | - |
0.864 | 5400 | 0.0001 | - |
0.872 | 5450 | 0.0001 | - |
0.88 | 5500 | 0.0001 | - |
0.888 | 5550 | 0.0001 | - |
0.896 | 5600 | 0.0 | - |
0.904 | 5650 | 0.0001 | - |
0.912 | 5700 | 0.0001 | - |
0.92 | 5750 | 0.0001 | - |
0.928 | 5800 | 0.0 | - |
0.936 | 5850 | 0.0 | - |
0.944 | 5900 | 0.0 | - |
0.952 | 5950 | 0.0 | - |
0.96 | 6000 | 0.0 | - |
0.968 | 6050 | 0.0 | - |
0.976 | 6100 | 0.0001 | - |
0.984 | 6150 | 0.0 | - |
0.992 | 6200 | 0.0 | - |
1.0 | 6250 | 0.0 | 0.3546 |
- The bold row denotes the saved checkpoint.
Framework Versions
- Python: 3.10.13
- SetFit: 1.0.3
- Sentence Transformers: 2.2.2
- Transformers: 4.36.2
- PyTorch: 2.1.2+cu121
- Datasets: 2.16.1
- Tokenizers: 0.15.0
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}
}