XLM_RoBERTa-Clickbait-Detection-new

<!-- 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_RoBERTa-Clickbait-Detection-new

This model is a fine-tuned version of xlm-roberta-large on the christinacdl/clickbait_detection_dataset dataset. It achieves the following results on the evaluation set:

• Loss: 0.1071
• Micro F1: 0.9834
• Macro F1: 0.9833
• Accuracy: 0.9834

It achieves the following results on the test set:

• Accuracy: 0.9838922630050172

• Micro-F1 Score: 0.9838922630050172

• Macro-F1 Score: 0.9838416247418498

• Matthews Correlation Coefficient: 0.9676867009951606

• Precision of each class: [0.98156425 0.98597897]

• Recall of each class: [0.98431373 0.98351648]

• F1 score of each class: [0.98293706 0.98474619]

Intended uses & limitations

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 4
• early stopping patience: 2
• adam epsilon: 1e-8
• gradient_checkpointing: True
• max_grad_norm: 1.0
• seed: 42
• optimizer: adamw_torch_fused
• weight decay: 0.01
• warmup_ratio: 0
• group_by_length: True
• max_seq_length: 512
• save_steps: 1000
• logging_steps: 500
• evaluation_strategy: epoch
• save_strategy: epoch
• eval_steps: 1000
• save_total_limit: 2

All results from Training and Evaluation

• "epoch": 4.0,
• "eval_accuracy": 0.9844203855294428,
• "eval_loss": 0.08027808368206024,
• "eval_macro_f1": 0.9843695357857132,
• "eval_micro_f1": 0.9844203855294428,
• "eval_runtime": 124.9733,
• "eval_samples": 3787,
• "eval_samples_per_second": 30.302,
• "eval_steps_per_second": 1.896,
• "predict_accuracy": 0.9838922630050172,
• "predict_loss": 0.07716809958219528,
• "predict_macro_f1": 0.9838416247418498,
• "predict_micro_f1": 0.9838922630050172,
• "predict_runtime": 127.7861,
• "predict_samples": 3787,
• "predict_samples_per_second": 29.635,
• "predict_steps_per_second": 1.855,
• "train_loss": 0.057462599486458765,
• "train_runtime": 25253.576,
• "train_samples": 30296,
• "train_samples_per_second": 4.799,
• "train_steps_per_second": 0.15

Framework versions

• Transformers 4.36.1
• Pytorch 2.1.0+cu121
• Datasets 2.13.1
• Tokenizers 0.15.0