finetuned_bart_for_titlekeygen_custom

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

• Loss: 4.9518
• Rouge1: 27.1931
• Rouge2: 17.9025
• Rougel: 27.0914
• Rougelsum: 27.5983
• Gen Len: 10.4444

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
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	0.91	5	4.4910	24.613	15.7516	23.7302	24.6083	11.0
No log	2.0	11	4.5644	23.1235	14.2857	22.4578	23.0522	9.0
No log	2.91	16	4.6286	24.7078	15.2652	24.7869	25.2108	10.0556
No log	4.0	22	4.6774	24.5206	17.2336	24.3437	25.1579	10.5
No log	4.91	27	4.7187	26.304	18.2158	26.3307	26.814	10.2778
No log	6.0	33	4.7962	26.7873	17.4056	26.8439	27.3369	10.1667
No log	6.91	38	4.8356	26.9721	17.3677	26.9926	27.5469	10.0556
No log	8.0	44	4.7903	28.0369	17.3677	28.1825	28.5073	10.0
No log	8.91	49	4.7801	27.1885	17.6215	26.6686	27.0771	10.1111
No log	10.0	55	4.8049	27.1885	17.6215	26.6686	27.0771	10.1667
No log	10.91	60	4.9275	26.8226	17.9025	26.8217	27.2015	10.0
No log	12.0	66	4.9744	26.8226	17.9025	26.8217	27.2015	10.1111
No log	12.91	71	4.9863	26.0231	17.9025	26.0376	26.3104	10.0
No log	14.0	77	4.9497	24.9154	17.2851	25.0693	25.3933	9.8333
No log	14.91	82	4.9272	24.5615	17.037	24.4986	25.0688	10.0556
No log	16.0	88	4.9417	25.307	17.037	25.3656	25.9459	10.0
No log	16.91	93	4.9712	26.7457	17.9025	26.7271	27.1185	9.9444
No log	18.0	99	4.9649	25.9581	17.9025	25.9478	26.2429	10.0556
No log	18.91	104	4.9305	27.0746	17.9025	26.9536	27.4569	10.6111
No log	20.0	110	4.9212	27.9706	18.5997	27.2022	27.7873	10.8889
No log	20.91	115	4.9196	27.7549	18.8225	27.5716	28.1145	10.6667
No log	22.0	121	4.9316	27.1931	17.9025	27.0914	27.5983	10.4444
No log	22.91	126	4.9525	27.1931	17.9025	27.0914	27.5983	10.2222
No log	24.0	132	4.9590	27.1931	17.9025	27.0914	27.5983	10.2222
No log	24.91	137	4.9545	27.1931	17.9025	27.0914	27.5983	10.2222
No log	26.0	143	4.9514	27.1931	17.9025	27.0914	27.5983	10.4444
No log	26.91	148	4.9515	27.1931	17.9025	27.0914	27.5983	10.4444
No log	27.27	150	4.9518	27.1931	17.9025	27.0914	27.5983	10.4444

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.2
• Datasets 2.1.0
• Tokenizers 0.15.1