t5-abs-1709-1203-lr-0.0001-bs-10-maxep-20

This model is a fine-tuned version of google-t5/t5-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.3944
• Rouge/rouge1: 0.4206
• Rouge/rouge2: 0.162
• Rouge/rougel: 0.33
• Rouge/rougelsum: 0.3303
• Bertscore/bertscore-precision: 0.8957
• Bertscore/bertscore-recall: 0.8758
• Bertscore/bertscore-f1: 0.8855
• Meteor: 0.3203
• Gen Len: 34.9

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: 0.0001
• train_batch_size: 10
• eval_batch_size: 10
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 20
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge/rouge1	Rouge/rouge2	Rouge/rougel	Rouge/rougelsum	Bertscore/bertscore-precision	Bertscore/bertscore-recall	Bertscore/bertscore-f1	Meteor	Gen Len
1.9289	0.8	2	2.0526	0.4186	0.188	0.3713	0.3708	0.9077	0.8772	0.8921	0.338	31.5
1.2808	2.0	5	2.0627	0.4127	0.1942	0.3695	0.3702	0.9053	0.8756	0.89	0.3277	30.6
1.7019	2.8	7	2.0780	0.4156	0.1796	0.3544	0.354	0.9035	0.8741	0.8884	0.3306	30.7
1.05	4.0	10	2.1041	0.3644	0.1342	0.3099	0.3086	0.8947	0.871	0.8825	0.2859	32.7
1.4533	4.8	12	2.1367	0.3557	0.12	0.2873	0.2883	0.8883	0.868	0.8779	0.2659	33.5
0.8959	6.0	15	2.2072	0.413	0.1484	0.3224	0.3221	0.8934	0.8744	0.8836	0.3015	34.3
1.2858	6.8	17	2.2363	0.3801	0.1284	0.2942	0.2948	0.8881	0.8696	0.8786	0.2635	34.8
0.819	8.0	20	2.2531	0.4042	0.1592	0.3155	0.3165	0.8941	0.8739	0.8837	0.3198	35.8
1.1514	8.8	22	2.2739	0.394	0.1581	0.3193	0.321	0.8944	0.8739	0.8838	0.312	33.0
0.7579	10.0	25	2.3118	0.4277	0.1766	0.3409	0.3425	0.8992	0.8767	0.8876	0.3405	35.2
1.1033	10.8	27	2.3370	0.422	0.1704	0.3423	0.3439	0.8972	0.8761	0.8864	0.3399	35.9
0.6976	12.0	30	2.3617	0.4205	0.1703	0.3281	0.3285	0.8974	0.8755	0.8861	0.3227	35.5
1.0212	12.8	32	2.3656	0.4179	0.1651	0.3234	0.3225	0.8974	0.8756	0.8861	0.3215	35.3
0.6757	14.0	35	2.3826	0.4187	0.1615	0.3282	0.3278	0.8962	0.8761	0.8859	0.32	35.2
1.0276	14.8	37	2.3884	0.4206	0.162	0.33	0.3303	0.8957	0.8758	0.8855	0.3203	34.9
0.6742	16.0	40	2.3944	0.4206	0.162	0.33	0.3303	0.8957	0.8758	0.8855	0.3203	34.9

Framework versions

• Transformers 4.44.0
• Pytorch 2.4.0
• Datasets 2.21.0
• Tokenizers 0.19.1