flan-t5-base-samsum

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

• Loss: 1.3859
• Rouge1: 47.1046
• Rouge2: 23.264
• Rougel: 39.2757
• Rougelsum: 43.2598
• Gen Len: 17.3333

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: 24
• eval_batch_size: 24
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
1.5121	0.08	50	1.4287	46.7868	22.863	38.971	42.8209	16.9634
1.46	0.16	100	1.4199	46.8031	22.8195	39.0708	42.8717	17.2393
1.4515	0.24	150	1.4147	46.6849	23.0376	38.9434	42.8344	17.1245
1.4679	0.33	200	1.4121	46.8756	22.8504	39.1671	43.1892	17.3431
1.451	0.41	250	1.4109	46.8572	23.09	39.2939	43.2955	17.2686
1.4434	0.49	300	1.4040	46.6829	23.071	39.3131	43.1432	16.9158
1.4417	0.57	350	1.4007	46.8637	23.0661	39.2462	43.1897	17.1172
1.4781	0.65	400	1.3952	46.8511	23.1134	39.3071	43.2164	17.2076
1.4626	0.73	450	1.3940	47.1533	23.2771	39.3094	43.2806	17.2222
1.4307	0.81	500	1.3955	46.9527	23.2227	39.2844	43.1903	17.2002
1.4586	0.9	550	1.3933	46.7523	23.1759	39.2675	43.1588	17.3040
1.4465	0.98	600	1.3905	46.855	23.3518	39.2879	43.2145	17.3468
1.381	1.06	650	1.3953	46.9719	22.9788	39.0886	43.1892	17.4066
1.4125	1.14	700	1.3922	46.535	23.0956	38.9275	42.9811	17.2381
1.3667	1.22	750	1.3922	47.3311	23.4123	39.5412	43.5624	17.2930
1.3878	1.3	800	1.3953	46.6737	23.2153	39.2982	43.2596	17.3358
1.3884	1.38	850	1.3931	46.9764	23.1561	39.1606	43.2115	17.3614
1.3766	1.47	900	1.3898	47.0466	23.1674	39.2822	43.293	17.3333
1.3727	1.55	950	1.3889	46.7311	23.0837	39.0882	43.0072	17.3211
1.4001	1.63	1000	1.3859	47.1046	23.264	39.2757	43.2598	17.3333
1.3894	1.71	1050	1.3874	47.2479	23.3762	39.4723	43.5241	17.3297
1.3697	1.79	1100	1.3860	47.1037	23.3894	39.3848	43.3875	17.3504
1.3886	1.87	1150	1.3862	47.0714	23.3937	39.4181	43.3841	17.3260
1.4037	1.95	1200	1.3861	47.0725	23.4085	39.3575	43.3676	17.3321

Framework versions

• Transformers 4.33.3
• Pytorch 2.0.0+cu117
• Datasets 2.14.5
• Tokenizers 0.13.3