flan-t5-small-samsum

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

• Loss: 1.6729
• Rouge1: 42.6
• Rouge2: 18.7153
• Rougel: 35.4138
• Rougelsum: 38.8543
• Gen Len: 16.9170

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: 5e-05
• train_batch_size: 32
• eval_batch_size: 32
• 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	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
1.8863	0.22	100	1.7049	42.0859	18.0002	34.7349	38.3446	16.5788
1.8463	0.43	200	1.6947	42.4056	18.3005	34.9821	38.8013	17.3614
1.8548	0.65	300	1.6792	42.585	18.5643	35.2235	38.8298	17.1514
1.8358	0.87	400	1.6772	42.1544	18.2303	34.8971	38.3609	16.5873
1.8129	1.08	500	1.6729	42.6	18.7153	35.4138	38.8543	16.9170
1.8068	1.3	600	1.6709	42.5217	18.3285	35.1455	38.5954	16.9451
1.7973	1.52	700	1.6687	42.8667	18.624	35.3429	38.9322	16.7546
1.7979	1.74	800	1.6668	42.919	18.7388	35.4528	39.0561	16.8791
1.7899	1.95	900	1.6670	43.0931	18.741	35.5047	39.2321	16.9109

Framework versions

• Transformers 4.36.0
• Pytorch 2.0.0
• Datasets 2.15.0
• Tokenizers 0.15.0