metadata
library_name: setfit
tags:
- setfit
- absa
- sentence-transformers
- text-classification
- generated_from_setfit_trainer
base_model: cointegrated/rubert-tiny2
metrics:
- accuracy
widget:
- text: Шеф - повар:Шеф - повар тоже с самого открытия .
- text: >-
ресторана:Сомнений по поводу выбора ресторана на свадьбу не возникло ,
надеюсь , что в самый важный день нашей жизни мы тоже останемся довольны ,
на этой неделе идем заказывать : ) .
- text: >-
гребешки:Затем были гребешки вроде ничего , но отдавали уксусом , пюре
вместе с ним было пересолено .
- text: >-
кафе:По кухне можно сказать , что это кафе для тех , кто любит соотношение
цены и качества .
- text: >-
то:Я не ходила в этот ресторан в детстве , не знаю , как всё было когда -
то , но сейчас это вполне симпатичное и уютное заведение с хорошей кухней
.
pipeline_tag: text-classification
inference: false
SetFit Aspect Model with cointegrated/rubert-tiny2
This is a SetFit model that can be used for Aspect Based Sentiment Analysis (ABSA). This SetFit model uses cointegrated/rubert-tiny2 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: cointegrated/rubert-tiny2
- Classification head: a LogisticRegression instance
- spaCy Model: en_core_web_lg
- SetFitABSA Aspect Model: isolation-forest/setfit-absa-aspect
- SetFitABSA Polarity Model: isolation-forest/setfit-absa-polarity
- Maximum Sequence Length: 2048 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 |
|
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(
"isolation-forest/setfit-absa-aspect",
"isolation-forest/setfit-absa-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 | 3 | 32.2987 | 171 |
Label | Training Sample Count |
---|---|
no aspect | 380 |
aspect | 256 |
Training Hyperparameters
- batch_size: (16, 2)
- 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: 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 |
---|---|---|---|
0.0001 | 1 | 0.2618 | - |
0.0038 | 50 | 0.2144 | - |
0.0076 | 100 | 0.2504 | - |
0.0114 | 150 | 0.2392 | - |
0.0152 | 200 | 0.2717 | - |
0.0190 | 250 | 0.2488 | - |
0.0228 | 300 | 0.2256 | - |
0.0266 | 350 | 0.2266 | - |
0.0304 | 400 | 0.2203 | - |
0.0342 | 450 | 0.2439 | - |
0.0380 | 500 | 0.2463 | - |
0.0418 | 550 | 0.3144 | - |
0.0456 | 600 | 0.1814 | - |
0.0494 | 650 | 0.1585 | - |
0.0532 | 700 | 0.0941 | - |
0.0570 | 750 | 0.1534 | - |
0.0608 | 800 | 0.0915 | - |
0.0646 | 850 | 0.1498 | - |
0.0684 | 900 | 0.0862 | - |
0.0722 | 950 | 0.0919 | - |
0.0760 | 1000 | 0.0252 | - |
0.0798 | 1050 | 0.0441 | - |
0.0836 | 1100 | 0.0808 | - |
0.0874 | 1150 | 0.1103 | - |
0.0912 | 1200 | 0.0138 | - |
0.0950 | 1250 | 0.052 | - |
0.0988 | 1300 | 0.0564 | - |
0.1026 | 1350 | 0.0058 | - |
0.1064 | 1400 | 0.0177 | - |
0.1102 | 1450 | 0.0651 | - |
0.1140 | 1500 | 0.0046 | - |
0.1178 | 1550 | 0.0046 | - |
0.1216 | 1600 | 0.0053 | - |
0.1254 | 1650 | 0.0464 | - |
0.1292 | 1700 | 0.0043 | - |
0.1330 | 1750 | 0.0403 | - |
0.1368 | 1800 | 0.0609 | - |
0.1406 | 1850 | 0.0093 | - |
0.1444 | 1900 | 0.0027 | - |
0.1482 | 1950 | 0.0041 | - |
0.1520 | 2000 | 0.0028 | - |
0.1558 | 2050 | 0.0072 | - |
0.1596 | 2100 | 0.0033 | - |
0.1634 | 2150 | 0.0029 | - |
0.1672 | 2200 | 0.0036 | - |
0.1710 | 2250 | 0.0019 | - |
0.1748 | 2300 | 0.0026 | - |
0.1786 | 2350 | 0.0544 | - |
0.1824 | 2400 | 0.0024 | - |
0.1862 | 2450 | 0.0028 | - |
0.1900 | 2500 | 0.0025 | - |
0.1938 | 2550 | 0.0018 | - |
0.1976 | 2600 | 0.0021 | - |
0.2014 | 2650 | 0.0023 | - |
0.2052 | 2700 | 0.0021 | - |
0.2090 | 2750 | 0.0026 | - |
0.2127 | 2800 | 0.0016 | - |
0.2165 | 2850 | 0.0023 | - |
0.2203 | 2900 | 0.0032 | - |
0.2241 | 2950 | 0.0019 | - |
0.2279 | 3000 | 0.0027 | - |
0.2317 | 3050 | 0.0035 | - |
0.2355 | 3100 | 0.0022 | - |
0.2393 | 3150 | 0.0019 | - |
0.2431 | 3200 | 0.0017 | - |
0.2469 | 3250 | 0.0016 | - |
0.2507 | 3300 | 0.0016 | - |
0.2545 | 3350 | 0.0017 | - |
0.2583 | 3400 | 0.0029 | - |
0.2621 | 3450 | 0.0017 | - |
0.2659 | 3500 | 0.0016 | - |
0.2697 | 3550 | 0.0019 | - |
0.2735 | 3600 | 0.0093 | - |
0.2773 | 3650 | 0.0023 | - |
0.2811 | 3700 | 0.0012 | - |
0.2849 | 3750 | 0.0016 | - |
0.2887 | 3800 | 0.0016 | - |
0.2925 | 3850 | 0.0021 | - |
0.2963 | 3900 | 0.0016 | - |
0.3001 | 3950 | 0.0017 | - |
0.3039 | 4000 | 0.0013 | - |
0.3077 | 4050 | 0.0017 | - |
0.3115 | 4100 | 0.0011 | - |
0.3153 | 4150 | 0.002 | - |
0.3191 | 4200 | 0.0015 | - |
0.3229 | 4250 | 0.001 | - |
0.3267 | 4300 | 0.0017 | - |
0.3305 | 4350 | 0.0011 | - |
0.3343 | 4400 | 0.0061 | - |
0.3381 | 4450 | 0.0057 | - |
0.3419 | 4500 | 0.0465 | - |
0.3457 | 4550 | 0.0016 | - |
0.3495 | 4600 | 0.0014 | - |
0.3533 | 4650 | 0.0013 | - |
0.3571 | 4700 | 0.0014 | - |
0.3609 | 4750 | 0.0018 | - |
0.3647 | 4800 | 0.0014 | - |
0.3685 | 4850 | 0.0013 | - |
0.3723 | 4900 | 0.0009 | - |
0.3761 | 4950 | 0.0008 | - |
0.3799 | 5000 | 0.0011 | - |
0.3837 | 5050 | 0.002 | - |
0.3875 | 5100 | 0.0014 | - |
0.3913 | 5150 | 0.001 | - |
0.3951 | 5200 | 0.0012 | - |
0.3989 | 5250 | 0.0017 | - |
0.4027 | 5300 | 0.0011 | - |
0.4065 | 5350 | 0.0012 | - |
0.4103 | 5400 | 0.0009 | - |
0.4141 | 5450 | 0.0015 | - |
0.4179 | 5500 | 0.0009 | - |
0.4217 | 5550 | 0.0012 | - |
0.4255 | 5600 | 0.0013 | - |
0.4293 | 5650 | 0.0465 | - |
0.4331 | 5700 | 0.0011 | - |
0.4369 | 5750 | 0.0008 | - |
0.4407 | 5800 | 0.0012 | - |
0.4445 | 5850 | 0.0008 | - |
0.4483 | 5900 | 0.0013 | - |
0.4521 | 5950 | 0.0011 | - |
0.4559 | 6000 | 0.0229 | - |
0.4597 | 6050 | 0.0012 | - |
0.4635 | 6100 | 0.0009 | - |
0.4673 | 6150 | 0.0011 | - |
0.4711 | 6200 | 0.0011 | - |
0.4749 | 6250 | 0.001 | - |
0.4787 | 6300 | 0.0008 | - |
0.4825 | 6350 | 0.0011 | - |
0.4863 | 6400 | 0.0012 | - |
0.4901 | 6450 | 0.0008 | - |
0.4939 | 6500 | 0.0014 | - |
0.4977 | 6550 | 0.001 | - |
0.5015 | 6600 | 0.0014 | - |
0.5053 | 6650 | 0.001 | - |
0.5091 | 6700 | 0.0008 | - |
0.5129 | 6750 | 0.0013 | - |
0.5167 | 6800 | 0.0012 | - |
0.5205 | 6850 | 0.0009 | - |
0.5243 | 6900 | 0.0008 | - |
0.5281 | 6950 | 0.001 | - |
0.5319 | 7000 | 0.0012 | - |
0.5357 | 7050 | 0.0009 | - |
0.5395 | 7100 | 0.0007 | - |
0.5433 | 7150 | 0.0008 | - |
0.5471 | 7200 | 0.001 | - |
0.5509 | 7250 | 0.0006 | - |
0.5547 | 7300 | 0.0007 | - |
0.5585 | 7350 | 0.0012 | - |
0.5623 | 7400 | 0.0159 | - |
0.5661 | 7450 | 0.0008 | - |
0.5699 | 7500 | 0.0012 | - |
0.5737 | 7550 | 0.0011 | - |
0.5775 | 7600 | 0.0008 | - |
0.5813 | 7650 | 0.0009 | - |
0.5851 | 7700 | 0.0005 | - |
0.5889 | 7750 | 0.0017 | - |
0.5927 | 7800 | 0.0009 | - |
0.5965 | 7850 | 0.0007 | - |
0.6003 | 7900 | 0.0065 | - |
0.6041 | 7950 | 0.0007 | - |
0.6079 | 8000 | 0.0041 | - |
0.6117 | 8050 | 0.0009 | - |
0.6155 | 8100 | 0.038 | - |
0.6193 | 8150 | 0.0005 | - |
0.6231 | 8200 | 0.0356 | - |
0.6269 | 8250 | 0.0007 | - |
0.6307 | 8300 | 0.0008 | - |
0.6345 | 8350 | 0.0009 | - |
0.6382 | 8400 | 0.001 | - |
0.6420 | 8450 | 0.0009 | - |
0.6458 | 8500 | 0.0008 | - |
0.6496 | 8550 | 0.0009 | - |
0.6534 | 8600 | 0.0009 | - |
0.6572 | 8650 | 0.0008 | - |
0.6610 | 8700 | 0.0006 | - |
0.6648 | 8750 | 0.0009 | - |
0.6686 | 8800 | 0.0006 | - |
0.6724 | 8850 | 0.0008 | - |
0.6762 | 8900 | 0.0008 | - |
0.6800 | 8950 | 0.0245 | - |
0.6838 | 9000 | 0.0007 | - |
0.6876 | 9050 | 0.0008 | - |
0.6914 | 9100 | 0.0007 | - |
0.6952 | 9150 | 0.0006 | - |
0.6990 | 9200 | 0.0009 | - |
0.7028 | 9250 | 0.0011 | - |
0.7066 | 9300 | 0.0009 | - |
0.7104 | 9350 | 0.0008 | - |
0.7142 | 9400 | 0.0008 | - |
0.7180 | 9450 | 0.0007 | - |
0.7218 | 9500 | 0.0006 | - |
0.7256 | 9550 | 0.0233 | - |
0.7294 | 9600 | 0.0008 | - |
0.7332 | 9650 | 0.0173 | - |
0.7370 | 9700 | 0.0006 | - |
0.7408 | 9750 | 0.0007 | - |
0.7446 | 9800 | 0.0007 | - |
0.7484 | 9850 | 0.001 | - |
0.7522 | 9900 | 0.0007 | - |
0.7560 | 9950 | 0.0006 | - |
0.7598 | 10000 | 0.0006 | - |
0.7636 | 10050 | 0.0008 | - |
0.7674 | 10100 | 0.0005 | - |
0.7712 | 10150 | 0.0007 | - |
0.7750 | 10200 | 0.0007 | - |
0.7788 | 10250 | 0.0009 | - |
0.7826 | 10300 | 0.0008 | - |
0.7864 | 10350 | 0.0007 | - |
0.7902 | 10400 | 0.0009 | - |
0.7940 | 10450 | 0.0007 | - |
0.7978 | 10500 | 0.0007 | - |
0.8016 | 10550 | 0.0008 | - |
0.8054 | 10600 | 0.0007 | - |
0.8092 | 10650 | 0.0007 | - |
0.8130 | 10700 | 0.0007 | - |
0.8168 | 10750 | 0.0007 | - |
0.8206 | 10800 | 0.0005 | - |
0.8244 | 10850 | 0.0007 | - |
0.8282 | 10900 | 0.0005 | - |
0.8320 | 10950 | 0.0005 | - |
0.8358 | 11000 | 0.0006 | - |
0.8396 | 11050 | 0.0008 | - |
0.8434 | 11100 | 0.0008 | - |
0.8472 | 11150 | 0.0137 | - |
0.8510 | 11200 | 0.0008 | - |
0.8548 | 11250 | 0.012 | - |
0.8586 | 11300 | 0.0006 | - |
0.8624 | 11350 | 0.0007 | - |
0.8662 | 11400 | 0.0007 | - |
0.8700 | 11450 | 0.0009 | - |
0.8738 | 11500 | 0.0007 | - |
0.8776 | 11550 | 0.0008 | - |
0.8814 | 11600 | 0.0005 | - |
0.8852 | 11650 | 0.0008 | - |
0.8890 | 11700 | 0.0008 | - |
0.8928 | 11750 | 0.0007 | - |
0.8966 | 11800 | 0.0006 | - |
0.9004 | 11850 | 0.0006 | - |
0.9042 | 11900 | 0.0006 | - |
0.9080 | 11950 | 0.0007 | - |
0.9118 | 12000 | 0.0005 | - |
0.9156 | 12050 | 0.0007 | - |
0.9194 | 12100 | 0.0006 | - |
0.9232 | 12150 | 0.0008 | - |
0.9270 | 12200 | 0.0006 | - |
0.9308 | 12250 | 0.0005 | - |
0.9346 | 12300 | 0.0167 | - |
0.9384 | 12350 | 0.0008 | - |
0.9422 | 12400 | 0.0005 | - |
0.9460 | 12450 | 0.0233 | - |
0.9498 | 12500 | 0.001 | - |
0.9536 | 12550 | 0.0006 | - |
0.9574 | 12600 | 0.0007 | - |
0.9612 | 12650 | 0.0007 | - |
0.9650 | 12700 | 0.0006 | - |
0.9688 | 12750 | 0.0008 | - |
0.9726 | 12800 | 0.0006 | - |
0.9764 | 12850 | 0.0177 | - |
0.9802 | 12900 | 0.0008 | - |
0.9840 | 12950 | 0.0007 | - |
0.9878 | 13000 | 0.0131 | - |
0.9916 | 13050 | 0.0007 | - |
0.9954 | 13100 | 0.0006 | - |
0.9992 | 13150 | 0.0004 | - |
Framework Versions
- Python: 3.10.13
- SetFit: 1.0.3
- Sentence Transformers: 2.6.1
- spaCy: 3.7.2
- Transformers: 4.39.3
- 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}
}