FLAN-T5-Base-Finetune-Remarks-Trend-V2

This model is a fine-tuned version of google/flan-t5-base on report card remarks dataset. It achieves the following results on the evaluation set:

• Loss: 0.5248
• Rouge1: 46.0468
• Rouge2: 27.6211
• Rougel: 40.1597
• Rougelsum: 40.1729
• Gen Len: 19.0

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: 4

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
0.8201	1.0	1180	0.6079	45.621	26.5385	39.602	39.6224	19.0
0.6879	2.0	2360	0.5507	45.6672	27.3069	40.0047	40.0248	19.0
0.63	3.0	3540	0.5315	45.9965	27.7033	40.1731	40.1997	19.0
0.618	4.0	4720	0.5248	46.0468	27.6211	40.1597	40.1729	19.0

Framework versions

• Transformers 4.44.0
• Pytorch 2.4.0
• Datasets 2.21.0
• Tokenizers 0.19.1