flan-t5-base-finetuned-QLoRA

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

• Loss: 1.0780
• Rouge1: 0.2435
• Rouge2: 0.1079
• Rougel: 0.1991
• Rougelsum: 0.2302

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 3e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
12.7942	1.0	250	10.7766	0.2346	0.1022	0.1834	0.2154
3.0774	2.0	500	2.5061	0.2351	0.1094	0.197	0.2204
2.1947	3.0	750	1.4702	0.2403	0.1104	0.1997	0.2261
1.7687	4.0	1000	1.2326	0.247	0.1148	0.2024	0.2307
1.4731	5.0	1250	1.1516	0.2538	0.1203	0.2074	0.2381
1.4802	6.0	1500	1.1120	0.2432	0.1102	0.1993	0.2271
1.3568	7.0	1750	1.0945	0.2427	0.1089	0.1991	0.2279
1.4054	8.0	2000	1.0843	0.2428	0.1076	0.1993	0.2293
1.3151	9.0	2250	1.0795	0.2432	0.1076	0.1991	0.2299
1.2669	10.0	2500	1.0780	0.2435	0.1079	0.1991	0.2302

Framework versions

• PEFT 0.8.2
• Transformers 4.37.0
• Pytorch 2.1.2
• Datasets 2.1.0
• Tokenizers 0.15.1