metadata
library_name: setfit
tags:
- setfit
- absa
- sentence-transformers
- text-classification
- generated_from_setfit_trainer
metrics:
- accuracy
widget:
- text: |-
Wifi:Go get your coffee on.
Hey the coffee was strong, so what else?
Easy FWY on/off access
Close the beach
Other food choice nearby
Smaller starbucks, with less seating both indoors and outdoors
Wifi was slow at the time
- text: >-
place:Stopped by after a long day visit to Santa Barbara. There were few
different places near by but remembering that 'smaller place with limited
menu' has 70/30 chance of being better than a big place with big menu.
It's generally a good ratio to keep in mind, and depending on the category
of food, like sushi, it leans to a higher ratio like 80/20.
Other places may have better seating and views, but I believe this place
has better food. Their clam chowder was the best I've had so far in
California (been here only for 5 months, I'm just getting started). Their
2 options of fresh oysters were good choices. Their lemonade was good and
sweet. I just wished their shrimp cocktail came with more ...Cocktail.
(Easy fix, just ask for more)
- text: >-
evening:Three reasons why it gets three stars:
1. The crab cakes were good and is a definitely must try!
2. The shrimp scampi was actually amazing in the sauce that it comes with,
so that's another must try!
3. The real reason why it is getting three stars is because service is
everything in ANY restaurant you go to. Service started off great,
waitress was attentive, but once we paid the bill and left a 20% tip, my
guests and I, which was only three of us, stayed at the table to finish
our drinks and we're looking at funny videos from a trip we went to. Point
is the waitress rudely told my friend to lower the volume on his phone,
yet other guests were just as loud and we were sitting OUTSIDE...where it
is already a loud environment!
I really want to give it 4 stars, but if I give 4 stars it changes it to,
"Yay! I'm a fan", but I am not. The only reason why it's not getting 1
star, is because the food was decent, the view is nice and also the
manager was extremely empathetic to the situation and it wasn't her fault
at all that her waitress was obviously having an off day. I have never met
a manager that attentive and she was incredible at handling and diffusing
the situation. I cannot thank her enough for salvaging the rest of our
evening for how poor the waitress treated paying customers.
- text: >-
Mediterranean:Pretty good food, just had a wrap and it was delicious
pretty much on Mediterranean or Greek style food around here. Petra's who
had really good Greek dinners closed
- text: >-
sauce:The chicken made worth the waiting, my mild sauce was awesome, the
honey mustard my favorite
pipeline_tag: text-classification
inference: false
base_model: sentence-transformers/all-MiniLM-L6-v2
model-index:
- name: SetFit Aspect Model 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: 0.9602649006622517
name: Accuracy
SetFit Aspect Model with sentence-transformers/all-MiniLM-L6-v2
This is a SetFit model that can be used for Aspect Based Sentiment Analysis (ABSA). This SetFit model uses sentence-transformers/all-MiniLM-L6-v2 as the Sentence Transformer embedding model. A LogisticRegression instance is used for classification. In particular, this model is in charge of filtering aspect span candidates.
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 this SetFit model to filter these possible aspect span candidates.
- Use a SetFit model to classify the filtered aspect span candidates.
Model Details
Model Description
- Model Type: SetFit
- Sentence Transformer body: sentence-transformers/all-MiniLM-L6-v2
- Classification head: a LogisticRegression instance
- spaCy Model: en_core_web_sm
- SetFitABSA Aspect Model: ginkgogo/setfit-absa-bge-small-en-v1.5-restaurants-aspect
- SetFitABSA Polarity Model: ginkgogo/setfit-absa-bge-small-en-v1.5-restaurants-polarity
- 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 |
---|---|
aspect |
|
no aspect |
|
Evaluation
Metrics
Label | Accuracy |
---|---|
all | 0.9603 |
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 🤗 Hub
model = AbsaModel.from_pretrained(
"ginkgogo/setfit-absa-bge-small-en-v1.5-restaurants-aspect",
"ginkgogo/setfit-absa-bge-small-en-v1.5-restaurants-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 | 21 | 200.4733 | 491 |
Label | Training Sample Count |
---|---|
no aspect | 411 |
aspect | 20 |
Training Hyperparameters
- batch_size: (50, 50)
- num_epochs: (5, 5)
- 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: True
Training Results
Epoch | Step | Training Loss | Validation Loss |
---|---|---|---|
0.0003 | 1 | 0.2706 | - |
0.0147 | 50 | 0.2856 | 0.3049 |
0.0294 | 100 | 0.2817 | 0.2904 |
0.0442 | 150 | 0.2453 | 0.2837 |
0.0589 | 200 | 0.2637 | 0.2756 |
0.0736 | 250 | 0.199 | 0.2668 |
0.0883 | 300 | 0.1917 | 0.2523 |
0.1031 | 350 | 0.1071 | 0.1889 |
0.1178 | 400 | 0.049 | 0.0826 |
0.1325 | 450 | 0.022 | 0.0718 |
0.1472 | 500 | 0.0275 | 0.0767 |
0.1620 | 550 | 0.0152 | 0.0779 |
0.1767 | 600 | 0.0185 | 0.0905 |
0.1914 | 650 | 0.0044 | 0.0785 |
0.2061 | 700 | 0.008 | 0.0896 |
- The bold row denotes the saved checkpoint.
Framework Versions
- Python: 3.10.12
- SetFit: 1.0.3
- Sentence Transformers: 2.6.0
- spaCy: 3.7.4
- Transformers: 4.39.1
- PyTorch: 2.2.1+cu121
- 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}
}