FLAN-T5-Base-Finetune-Remarks-Bar

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

• Loss: 0.6775
• Rouge1: 38.4803
• Rouge2: 22.0178
• Rougel: 31.4848
• Rougelsum: 31.4757
• 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
0.9766	1.0	1125	0.7627	38.7818	21.7574	31.4008	31.3835	19.0
0.864	2.0	2250	0.7055	38.2489	21.7373	31.1157	31.1052	19.0
0.8192	3.0	3375	0.6831	38.7844	22.1673	31.5798	31.5805	19.0
0.7974	4.0	4500	0.6775	38.4803	22.0178	31.4848	31.4757	19.0

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0
• Datasets 3.0.0
• Tokenizers 0.19.1