bart-finetuned-idl

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: 0.0031
• Bleu: 0.0
• Gen Len: 4.9917

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: 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: 35
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
0.2005	1.0	13874	0.1589	0.0	5.0002
0.1182	2.0	27748	0.0949	0.0	4.9924
0.0983	3.0	41622	0.0778	0.0	4.9924
0.0724	4.0	55496	0.0724	0.0	4.9903
0.0532	5.0	69370	0.0549	0.0	4.9928
0.0458	6.0	83244	0.0463	0.0	4.9861
0.0435	7.0	97118	0.0548	0.0	4.9923
0.0464	8.0	110992	0.0847	0.0	4.9899
0.0317	9.0	124866	0.0303	0.0	4.9922
0.0302	10.0	138740	0.0284	0.0	4.9919
0.0306	11.0	152614	0.0120	0.0	4.9919
0.0224	12.0	166488	0.0462	0.0	4.9917
0.0184	13.0	180362	0.0138	0.0	4.9924
0.0208	14.0	194236	0.0730	0.0	4.9919
0.0149	15.0	208110	0.0126	0.0	4.992
0.0161	16.0	221984	0.0100	0.0	4.9915
0.0178	17.0	235858	0.0106	0.0	4.992
0.0116	18.0	249732	0.0149	0.0	4.9921
0.0096	19.0	263606	0.0085	0.0	4.9918
0.0094	20.0	277480	0.0101	0.0	4.9916
0.0084	21.0	291354	0.0093	0.0	4.9918
0.0077	22.0	305228	0.0138	0.0	4.992
0.0094	23.0	319102	0.0084	0.0	4.9918
0.0079	24.0	332976	0.0058	0.0	4.9917
0.006	25.0	346850	0.0067	0.0	4.9918
0.0046	26.0	360724	0.0041	0.0	4.9918
0.0049	27.0	374598	0.0061	0.0	4.9919
0.002	28.0	388472	0.0035	0.0	4.9918
0.003	29.0	402346	0.0038	0.0	4.9917
0.0027	30.0	416220	0.0050	0.0	4.9917
0.001	31.0	430094	0.0063	0.0	4.9918
0.0017	32.0	443968	0.0042	0.0	4.992
0.0013	33.0	457842	0.0032	0.0	4.9917
0.0005	34.0	471716	0.0031	0.0	4.9917
0.0003	35.0	485590	0.0031	0.0	4.9917

Framework versions

• Transformers 4.24.0
• Pytorch 1.10.0+cu111
• Datasets 2.7.1
• Tokenizers 0.13.2