wikihow_t5small_LoRA_fine_tuned

This model is a fine-tuned version of t5-small on the wikihow dataset. It achieves the following results on the evaluation set:

• Loss: 0.2557
• BLEU: 0.34
• Rouge1: 0.33
• Rouge2: 0.065
• RougeL: 0.18

Intended uses & limitations

This work is the final project of an academic Natural Language Processing course

Training and evaluation data

wikiHow data consisting of the below details:

• Training data: 20000 datapoints
• Evaluation data: 5577 datapoints

Data has below columns:

• text: actual text of the article
• headline: concatenation of all the article headlines

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 2
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.4494	1.0	1250	0.2684
0.2821	2.0	2500	0.2557

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0