fine-tuned-bart-20-epochs

This model is a fine-tuned version of facebook/bart-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8098
• Rouge1: 0.3246
• Rouge2: 0.1287
• Rougel: 0.2921
• Rougelsum: 0.2912
• Gen Len: 14.96

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: 2e-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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	301	0.7895	0.2498	0.0834	0.2162	0.2159	14.58
1.8122	2.0	602	0.7331	0.2226	0.0794	0.1943	0.1931	13.51
1.8122	3.0	903	0.7235	0.2935	0.1013	0.2652	0.2647	14.69
0.6848	4.0	1204	0.7225	0.322	0.1245	0.2867	0.2857	13.92
0.5826	5.0	1505	0.7238	0.322	0.1149	0.2863	0.2854	14.81
0.5826	6.0	1806	0.7204	0.3255	0.1212	0.2977	0.2963	14.98
0.5013	7.0	2107	0.7377	0.3061	0.1104	0.2784	0.2767	14.84
0.5013	8.0	2408	0.7396	0.3092	0.1227	0.275	0.2741	14.17
0.4384	9.0	2709	0.7413	0.3224	0.1271	0.2935	0.2928	14.44
0.3952	10.0	3010	0.7458	0.3288	0.1302	0.2925	0.2925	15.09
0.3952	11.0	3311	0.7615	0.3496	0.139	0.3139	0.3137	15.13
0.3626	12.0	3612	0.7733	0.3311	0.1264	0.3057	0.3049	14.84
0.3626	13.0	3913	0.7779	0.3184	0.1226	0.286	0.2857	15.02
0.3254	14.0	4214	0.7854	0.3258	0.1199	0.2911	0.2915	14.89
0.2983	15.0	4515	0.7863	0.3346	0.1189	0.3027	0.3009	14.93
0.2983	16.0	4816	0.7979	0.3201	0.117	0.2857	0.2843	15.05
0.2807	17.0	5117	0.8037	0.3223	0.1216	0.291	0.2899	15.1
0.2807	18.0	5418	0.8048	0.3313	0.1261	0.3003	0.2996	15.1
0.2653	19.0	5719	0.8114	0.3285	0.1298	0.297	0.2963	15.01
0.2562	20.0	6020	0.8098	0.3246	0.1287	0.2921	0.2912	14.96

Framework versions

• Transformers 4.36.2
• Pytorch 1.12.1+cu113
• Datasets 2.15.0
• Tokenizers 0.15.0