SetFit Polarity Model
This is a SetFit model that can be used for Aspect Based Sentiment Analysis (ABSA). A LogisticRegression instance is used for classification. In particular, this model is in charge of classifying aspect polarities.
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.
This model was trained within the context of a larger system for ABSA, which looks like so:
- Use a spaCy model to select possible aspect span candidates.
- Use a SetFit model to filter these possible aspect span candidates.
- Use this SetFit model to classify the filtered aspect span candidates.
Model Details
Model Description
- Model Type: SetFit
- Classification head: a LogisticRegression instance
- spaCy Model: id_core_news_trf
- SetFitABSA Aspect Model: zeroix07/indo-setfit-absa-model-aspect
- SetFitABSA Polarity Model: zeroix07/indo-setfit-absa-model-polarity
- Maximum Sequence Length: 8192 tokens
- Number of Classes: 3 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 |
---|---|
positif |
|
netral |
|
negatif |
|
Evaluation
Metrics
Label | Accuracy |
---|---|
all | 0.6569 |
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 AbsaModel
# Download from the ๐ค Hub
model = AbsaModel.from_pretrained(
"zeroix07/indo-setfit-absa-model-aspect",
"zeroix07/indo-setfit-absa-model-polarity",
)
# Run inference
preds = model("The food was great, but the venue is just way too busy.")
Training Details
Training Set Metrics
Training set | Min | Median | Max |
---|---|---|---|
Word count | 5 | 21.6519 | 45 |
Label | Training Sample Count |
---|---|
konflik | 0 |
negatif | 48 |
netral | 69 |
positif | 64 |
Training Hyperparameters
- batch_size: (6, 6)
- num_epochs: (1, 16)
- max_steps: -1
- sampling_strategy: oversampling
- body_learning_rate: (2e-05, 1e-05)
- head_learning_rate: 0.01
- loss: CosineSimilarityLoss
- distance_metric: cosine_distance
- margin: 0.25
- end_to_end: False
- use_amp: True
- warmup_proportion: 0.1
- seed: 42
- eval_max_steps: -1
- load_best_model_at_end: False
Training Results
Epoch | Step | Training Loss | Validation Loss |
---|---|---|---|
0.0003 | 1 | 0.2985 | - |
0.0139 | 50 | 0.14 | - |
0.0278 | 100 | 0.0913 | - |
0.0417 | 150 | 0.0447 | - |
0.0556 | 200 | 0.0932 | - |
0.0694 | 250 | 0.2864 | - |
0.0833 | 300 | 0.2556 | - |
0.0972 | 350 | 0.1447 | - |
0.1111 | 400 | 0.0084 | - |
0.125 | 450 | 0.003 | - |
0.1389 | 500 | 0.0035 | - |
0.1528 | 550 | 0.0074 | - |
0.1667 | 600 | 0.0031 | - |
0.1806 | 650 | 0.0014 | - |
0.1944 | 700 | 0.002 | - |
0.2083 | 750 | 0.0006 | - |
0.2222 | 800 | 0.0005 | - |
0.2361 | 850 | 0.0005 | - |
0.25 | 900 | 0.0005 | - |
0.2639 | 950 | 0.0015 | - |
0.2778 | 1000 | 0.0007 | - |
0.2917 | 1050 | 0.0006 | - |
0.3056 | 1100 | 0.0006 | - |
0.3194 | 1150 | 0.0007 | - |
0.3333 | 1200 | 0.0091 | - |
0.3472 | 1250 | 0.0004 | - |
0.3611 | 1300 | 0.0003 | - |
0.375 | 1350 | 0.0005 | - |
0.3889 | 1400 | 0.0006 | - |
0.4028 | 1450 | 0.0434 | - |
0.4167 | 1500 | 0.0006 | - |
0.4306 | 1550 | 0.0003 | - |
0.4444 | 1600 | 0.0005 | - |
0.4583 | 1650 | 0.0004 | - |
0.4722 | 1700 | 0.0021 | - |
0.4861 | 1750 | 0.0012 | - |
0.5 | 1800 | 0.0004 | - |
0.5139 | 1850 | 0.0005 | - |
0.5278 | 1900 | 0.0004 | - |
0.5417 | 1950 | 0.0003 | - |
0.5556 | 2000 | 0.0003 | - |
0.5694 | 2050 | 0.0005 | - |
0.5833 | 2100 | 0.0004 | - |
0.5972 | 2150 | 0.0004 | - |
0.6111 | 2200 | 0.0005 | - |
0.625 | 2250 | 0.0004 | - |
0.6389 | 2300 | 0.0005 | - |
0.6528 | 2350 | 0.0004 | - |
0.6667 | 2400 | 0.0003 | - |
0.6806 | 2450 | 0.0004 | - |
0.6944 | 2500 | 0.0007 | - |
0.7083 | 2550 | 0.0003 | - |
0.7222 | 2600 | 0.0003 | - |
0.7361 | 2650 | 0.101 | - |
0.75 | 2700 | 0.0003 | - |
0.7639 | 2750 | 0.0004 | - |
0.7778 | 2800 | 0.0004 | - |
0.7917 | 2850 | 0.0003 | - |
0.8056 | 2900 | 0.0004 | - |
0.8194 | 2950 | 0.0899 | - |
0.8333 | 3000 | 0.0003 | - |
0.8472 | 3050 | 0.0002 | - |
0.8611 | 3100 | 0.0002 | - |
0.875 | 3150 | 0.0003 | - |
0.8889 | 3200 | 0.0002 | - |
0.9028 | 3250 | 0.0003 | - |
0.9167 | 3300 | 0.0004 | - |
0.9306 | 3350 | 0.0003 | - |
0.9444 | 3400 | 0.0003 | - |
0.9583 | 3450 | 0.0547 | - |
0.9722 | 3500 | 0.0003 | - |
0.9861 | 3550 | 0.0004 | - |
1.0 | 3600 | 0.0002 | - |
Framework Versions
- Python: 3.10.13
- SetFit: 1.0.3
- Sentence Transformers: 2.7.0
- spaCy: 3.7.4
- Transformers: 4.36.2
- PyTorch: 2.1.2
- Datasets: 2.18.0
- 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}
}
- Downloads last month
- 31
Inference API (serverless) has been turned off for this model.