sarcasm-detection

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

xlm-r-base-finetuned-indo-sarcasm

This model is a fine-tuned version of xlm-roberta-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5255
• Accuracy: 0.7957
• Recall: 0.6176
• Precision: 0.5868
• F1: 0.6018

Model description

This model is a test run for finetuning XLM-RoBERTa on a non-english sarcasm dataset

Intended uses & limitations

This is for school project purposes only (project presentation)

Training and evaluation data

The dataset used is a dataset intended for sarcasm detection in the Indonesian language. The dataset is split into three sets: test, validate, training.

Training procedure

https://colab.research.google.com/github/huggingface/workshops/blob/main/luzern-university/02-text-classification.ipynb#scrollTo=7a2420dd-b6ca-4c41-811d-e4922e400339 This was followed as a guide, as well as the NLP course provided by the Hugging face.

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1