--- library_name: setfit tags: - setfit - absa - sentence-transformers - text-classification - generated_from_setfit_trainer base_model: cointegrated/rubert-tiny2 metrics: - accuracy widget: - text: Посетили вчера Твинс с подругой ,:Посетили вчера Твинс с подругой , в целом все очень понравилось ! ! - text: ', что это кафе для тех ,:По кухне можно сказать , что это кафе для тех , кто любит соотношение цены и качества .' - text: особенно шашлыки и наполеон . ( спасибо:Готовят очень вкусно , особенно шашлыки и наполеон . ( спасибо большое поварам ) - text: 'свет , ненавязчивая музыка ( даже как:Интерьер приятный : есть гардероб , диваны , приглушенный свет , ненавязчивая музыка ( даже как - то раз наткнулись там на саксофониста ) , приятная атмосфера . . .' - text: 'отдельно : есть официанты , которые работают:По обслуживание отдельно : есть официанты , которые работают с самого открытия - это тоже неплохой показатель качества .' pipeline_tag: text-classification inference: false --- # SetFit Polarity Model with cointegrated/rubert-tiny2 This is a [SetFit](https://github.com/huggingface/setfit) model that can be used for Aspect Based Sentiment Analysis (ABSA). This SetFit model uses [cointegrated/rubert-tiny2](https://huggingface.co/cointegrated/rubert-tiny2) as the Sentence Transformer embedding model. A [LogisticRegression](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html) 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: 1. Fine-tuning a [Sentence Transformer](https://www.sbert.net) with contrastive learning. 2. 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: 1. Use a spaCy model to select possible aspect span candidates. 2. Use a SetFit model to filter these possible aspect span candidates. 3. **Use this SetFit model to classify the filtered aspect span candidates.** ## Model Details ### Model Description - **Model Type:** SetFit - **Sentence Transformer body:** [cointegrated/rubert-tiny2](https://huggingface.co/cointegrated/rubert-tiny2) - **Classification head:** a [LogisticRegression](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html) instance - **spaCy Model:** en_core_web_lg - **SetFitABSA Aspect Model:** [isolation-forest/setfit-absa-aspect](https://huggingface.co/isolation-forest/setfit-absa-aspect) - **SetFitABSA Polarity Model:** [isolation-forest/setfit-absa-polarity](https://huggingface.co/isolation-forest/setfit-absa-polarity) - **Maximum Sequence Length:** 2048 tokens - **Number of Classes:** 3 classes ### Model Sources - **Repository:** [SetFit on GitHub](https://github.com/huggingface/setfit) - **Paper:** [Efficient Few-Shot Learning Without Prompts](https://arxiv.org/abs/2209.11055) - **Blogpost:** [SetFit: Efficient Few-Shot Learning Without Prompts](https://huggingface.co/blog/setfit) ### Model Labels | Label | Examples | |:---------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| | Positive | | | Neutral | | | Negative | | ## Uses ### Direct Use for Inference First install the SetFit library: ```bash pip install setfit ``` Then you can load this model and run inference. ```python 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 | 5 | 27.2578 | 171 | | Label | Training Sample Count | |:---------|:----------------------| | Negative | 54 | | Neutral | 19 | | Positive | 183 | ### 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 | |:------:|:----:|:-------------:|:---------------:| | 0.0004 | 1 | 0.2493 | - | | 0.0216 | 50 | 0.2343 | - | | 0.0432 | 100 | 0.2786 | - | | 0.0648 | 150 | 0.1976 | - | | 0.0864 | 200 | 0.2 | - | | 0.1080 | 250 | 0.1894 | - | | 0.1296 | 300 | 0.081 | - | | 0.1513 | 350 | 0.1189 | - | | 0.1729 | 400 | 0.0279 | - | | 0.1945 | 450 | 0.0755 | - | | 0.2161 | 500 | 0.0436 | - | | 0.2377 | 550 | 0.0231 | - | | 0.2593 | 600 | 0.0088 | - | | 0.2809 | 650 | 0.0686 | - | | 0.3025 | 700 | 0.0138 | - | | 0.3241 | 750 | 0.0137 | - | | 0.3457 | 800 | 0.0087 | - | | 0.3673 | 850 | 0.0131 | - | | 0.3889 | 900 | 0.0245 | - | | 0.4105 | 950 | 0.0093 | - | | 0.4322 | 1000 | 0.0036 | - | | 0.4538 | 1050 | 0.0149 | - | | 0.4754 | 1100 | 0.02 | - | | 0.4970 | 1150 | 0.0387 | - | | 0.5186 | 1200 | 0.017 | - | | 0.5402 | 1250 | 0.0417 | - | | 0.5618 | 1300 | 0.0041 | - | | 0.5834 | 1350 | 0.0041 | - | | 0.6050 | 1400 | 0.0282 | - | | 0.6266 | 1450 | 0.0102 | - | | 0.6482 | 1500 | 0.0037 | - | | 0.6698 | 1550 | 0.0058 | - | | 0.6914 | 1600 | 0.0078 | - | | 0.7131 | 1650 | 0.0272 | - | | 0.7347 | 1700 | 0.0224 | - | | 0.7563 | 1750 | 0.0057 | - | | 0.7779 | 1800 | 0.0026 | - | | 0.7995 | 1850 | 0.0088 | - | | 0.8211 | 1900 | 0.0044 | - | | 0.8427 | 1950 | 0.005 | - | | 0.8643 | 2000 | 0.0026 | - | | 0.8859 | 2050 | 0.0072 | - | | 0.9075 | 2100 | 0.0033 | - | | 0.9291 | 2150 | 0.0047 | - | | 0.9507 | 2200 | 0.0048 | - | | 0.9723 | 2250 | 0.0042 | - | | 0.9939 | 2300 | 0.0043 | - | ### 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 ```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} } ```