bart-base-finetuned-question-to-answer

This model is a fine-tuned version of facebook/bart-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0108
• Bleu: 59.1046
• Gen Len: 20.0

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
2.559	1.0	516	0.6275	5.9858	18.5
2.3757	2.0	1032	0.5381	6.9001	18.4
2.1938	3.0	1548	0.5004	9.2119	17.8
2.0192	4.0	2064	0.3875	19.4081	18.1
1.8823	5.0	2580	0.3911	23.1245	18.2
1.7773	6.0	3096	0.3360	35.602	19.1
1.6561	7.0	3612	0.3018	22.9533	17.6
1.5494	8.0	4128	0.2970	32.4812	19.2
1.4596	9.0	4644	0.2351	46.2092	19.9
1.3723	10.0	5160	0.2382	42.1352	19.4
1.3056	11.0	5676	0.2203	43.1825	19.5
1.2302	12.0	6192	0.2005	38.4359	19.4
1.1611	13.0	6708	0.1694	43.435	19.5
1.0921	14.0	7224	0.1600	46.2221	19.5
1.0521	15.0	7740	0.1365	43.6428	19.5
0.9797	16.0	8256	0.1229	47.1793	19.5
0.9153	17.0	8772	0.1048	53.0445	20.0
0.8932	18.0	9288	0.1171	53.0445	20.0
0.8507	19.0	9804	0.0954	48.5863	18.9
0.7885	20.0	10320	0.0794	53.5876	19.0
0.7645	21.0	10836	0.0769	52.3334	18.9
0.7204	22.0	11352	0.0701	48.3328	18.9
0.685	23.0	11868	0.0576	52.7649	19.2
0.6524	24.0	12384	0.0521	54.2149	19.4
0.6302	25.0	12900	0.0486	54.2149	19.4
0.5926	26.0	13416	0.0408	59.1046	20.0
0.5701	27.0	13932	0.0419	48.1233	18.8
0.5483	28.0	14448	0.0418	54.0129	19.3
0.5271	29.0	14964	0.0314	59.1046	20.0
0.501	30.0	15480	0.0283	59.1046	20.0
0.4821	31.0	15996	0.0316	57.5434	19.7
0.4474	32.0	16512	0.0296	57.5434	19.7
0.4328	33.0	17028	0.0229	57.2181	19.7
0.4171	34.0	17544	0.0212	57.5434	19.7
0.4051	35.0	18060	0.0194	59.1046	20.0
0.3924	36.0	18576	0.0161	59.1046	20.0
0.3783	37.0	19092	0.0155	59.1046	20.0
0.3695	38.0	19608	0.0149	59.1046	20.0
0.3626	39.0	20124	0.0140	59.1046	20.0
0.3492	40.0	20640	0.0147	59.1046	20.0
0.3446	41.0	21156	0.0140	59.1046	20.0
0.3377	42.0	21672	0.0125	59.1046	20.0
0.3265	43.0	22188	0.0122	59.1046	20.0
0.3213	44.0	22704	0.0118	59.1046	20.0
0.3154	45.0	23220	0.0116	59.1046	20.0
0.3146	46.0	23736	0.0113	59.1046	20.0
0.3077	47.0	24252	0.0107	59.1046	20.0
0.304	48.0	24768	0.0109	59.1046	20.0
0.3063	49.0	25284	0.0107	59.1046	20.0
0.2998	50.0	25800	0.0108	59.1046	20.0

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0