| language: en | |
| tags: | |
| - text-classification | |
| - pytorch | |
| - roberta | |
| - emotions | |
| - multi-class-classification | |
| - multi-label-classification | |
| datasets: | |
| - go_emotions | |
| license: mit | |
| widget: | |
| - text: "I am not having a great day." | |
| Model trained from [roberta-base](https://huggingface.co/roberta-base) on the [go_emotions](https://huggingface.co/datasets/go_emotions) dataset for multi-label classification. | |
| [go_emotions](https://huggingface.co/datasets/go_emotions) is based on Reddit data and has 28 labels. It is a multi-label dataset where one or multiple labels may apply for any given input text, hence this model is a multi-label classification model with 28 'probability' float outputs for any given input text. Typically a threshold of 0.5 is applied to the probabilities for the prediction for each label. | |
| The model was trained using `AutoModelForSequenceClassification.from_pretrained` with `problem_type="multi_label_classification"` for 3 epochs with a learning rate of 2e-5 and weight decay of 0.01. | |
| Evaluation (of the 28 dim output via a threshold of 0.5 to binarize each) using the dataset test split gives: | |
| - Micro F1 0.585 | |
| - ROC AUC 0.751 | |
| - Accuracy 0.474 | |
| But the metrics would be more meaningful when measured per label given the multi-label nature. | |
| Additionally some labels (E.g. `gratitude`) when considered independently perform very strongly with F1 around 0.9, whilst others (E.g. `relief`) perform very poorly. This is a challenging dataset. Labels such as `relief` do have much fewer examples in the training data (less than 100 out of the 40k+), but there is also some ambiguity and/or labelling errors visible in the training data of `go_emotions` that is suspected to constrain the performance. |