t5_long_27-03-2024_14-47-48

This model is a fine-tuned version of Stancld/longt5-tglobal-large-16384-pubmed-3k_steps on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3435
• Rouge1: 18.0761
• Rouge2: 6.2848
• Rougel: 15.9805
• Rougelsum: 16.9344

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: 0.0005
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
18.5106	0.01	1	2.2998	9.0413	1.7951	7.3243	8.0357
8.5122	0.01	2	0.5275	0.0	0.0	0.0	0.0
10.8448	0.02	3	0.6630	0.0	0.0	0.0	0.0
9.4742	0.03	4	0.6472	0.0	0.0	0.0	0.0
8.8401	0.03	5	0.5541	0.0	0.0	0.0	0.0
5.2721	0.04	6	0.5268	0.0	0.0	0.0	0.0
7.4134	0.05	7	0.5197	0.0	0.0	0.0	0.0
8.1236	0.05	8	0.5084	0.0	0.0	0.0	0.0
4.9603	0.06	9	0.4955	0.0	0.0	0.0	0.0
7.9812	0.07	10	0.4838	0.0	0.0	0.0	0.0
10.1557	0.07	11	0.4729	0.0	0.0	0.0	0.0
9.943	0.08	12	0.4623	0.0	0.0	0.0	0.0
8.4195	0.09	13	0.4515	0.0	0.0	0.0	0.0
9.0108	0.09	14	0.4419	0.0	0.0	0.0	0.0
7.8627	0.1	15	0.4339	0.0	0.0	0.0	0.0
8.1388	0.11	16	0.4271	0.0	0.0	0.0	0.0
7.8132	0.11	17	0.4223	0.0	0.0	0.0	0.0
6.083	0.12	18	0.4186	0.0	0.0	0.0	0.0
11.2734	0.13	19	0.4137	0.0	0.0	0.0	0.0
6.004	0.13	20	0.4082	0.0	0.0	0.0	0.0
8.7849	0.14	21	0.4019	0.0	0.0	0.0	0.0
5.829	0.15	22	0.3976	0.0	0.0	0.0	0.0
7.0927	0.15	23	0.3929	0.0	0.0	0.0	0.0
7.5678	0.16	24	0.3887	0.0	0.0	0.0	0.0
7.1876	0.17	25	0.3848	0.0	0.0	0.0	0.0
8.5662	0.17	26	0.3824	0.0	0.0	0.0	0.0
9.3966	0.18	27	0.3804	0.0	0.0	0.0	0.0
10.1809	0.19	28	0.3780	0.0	0.0	0.0	0.0
8.0879	0.19	29	0.3765	0.0	0.0	0.0	0.0
7.5633	0.2	30	0.3749	0.0	0.0	0.0	0.0
7.277	0.21	31	0.3738	0.0	0.0	0.0	0.0
6.6679	0.21	32	0.3710	0.0	0.0	0.0	0.0
11.7409	0.22	33	0.3680	0.0	0.0	0.0	0.0
7.5637	0.23	34	0.3658	0.0	0.0	0.0	0.0
9.0556	0.23	35	0.3623	0.0	0.0	0.0	0.0
9.5907	0.24	36	0.3615	0.0	0.0	0.0	0.0
9.0023	0.25	37	0.3604	0.0	0.0	0.0	0.0
7.8242	0.25	38	0.3599	0.0	0.0	0.0	0.0
9.2029	0.26	39	0.3591	0.0	0.0	0.0	0.0
7.6971	0.27	40	0.3566	0.0	0.0	0.0	0.0
8.4237	0.27	41	0.3542	0.0	0.0	0.0	0.0
6.9863	0.28	42	0.3521	0.0	0.0	0.0	0.0
6.7574	0.29	43	0.3503	0.0	0.0	0.0	0.0
5.1329	0.29	44	0.3490	0.0	0.0	0.0	0.0
9.0936	0.3	45	0.3483	1.3799	0.4852	1.0924	1.2108
9.8391	0.31	46	0.3475	13.0878	4.5581	11.0376	11.8991
6.4842	0.31	47	0.3463	17.4226	6.4078	15.3093	16.2477
6.4921	0.32	48	0.3452	18.5772	6.8275	16.3128	17.2935
10.4488	0.33	49	0.3443	18.2004	6.5001	16.0497	17.0731
4.3364	0.33	50	0.3435	18.0761	6.2848	15.9805	16.9344
9.4075	0.34	51	0.3427	18.3684	6.4989	16.2721	17.2338
13.213	0.35	52	0.3423	18.1592	6.1019	16.0076	17.061
8.5205	0.35	53	0.3420	17.6529	5.8026	15.4966	16.4479
8.6332	0.36	54	0.3411	18.1603	6.1679	15.9369	16.9204
8.288	0.37	55	0.3404	18.3122	6.1727	15.9244	16.9652

Framework versions

• PEFT 0.10.0
• Transformers 4.37.2
• Pytorch 2.2.0+cu121
• Datasets 2.17.0
• Tokenizers 0.15.2