dit-tiny_rvl_cdip_100_examples_per_class_simkd_CEKD_t1_aNone

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

• Loss: 0.1502
• Accuracy: 0.0625
• Brier Loss: 0.9374
• Nll: 9.1398
• F1 Micro: 0.0625
• F1 Macro: 0.0074
• Ece: 0.1015
• Aurc: 0.9383

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: 16
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Brier Loss	Nll	F1 Micro	F1 Macro	Ece	Aurc
No log	0.96	12	0.1540	0.0625	0.9376	8.5438	0.0625	0.0074	0.1043	0.9530
No log	1.96	24	0.1519	0.0625	0.9376	8.2831	0.0625	0.0074	0.1008	0.9465
No log	2.96	36	0.1512	0.0625	0.9375	8.4629	0.0625	0.0074	0.1028	0.9336
No log	3.96	48	0.1510	0.0625	0.9375	8.6283	0.0625	0.0074	0.1027	0.9365
No log	4.96	60	0.1509	0.0625	0.9375	8.5065	0.0625	0.0074	0.1030	0.9433
No log	5.96	72	0.1508	0.0625	0.9375	8.4779	0.0625	0.0074	0.1017	0.9414
No log	6.96	84	0.1507	0.0625	0.9375	8.5053	0.0625	0.0074	0.1045	0.9438
No log	7.96	96	0.1507	0.0625	0.9375	8.7396	0.0625	0.0074	0.1032	0.9440
No log	8.96	108	0.1506	0.0625	0.9375	8.6420	0.0625	0.0074	0.1031	0.9448
No log	9.96	120	0.1506	0.0625	0.9375	8.8410	0.0625	0.0074	0.1045	0.9438
No log	10.96	132	0.1506	0.0625	0.9374	8.9438	0.0625	0.0074	0.1042	0.9413
No log	11.96	144	0.1505	0.0625	0.9374	8.9847	0.0625	0.0074	0.1032	0.9418
No log	12.96	156	0.1505	0.0625	0.9374	9.0594	0.0625	0.0074	0.1031	0.9397
No log	13.96	168	0.1504	0.0625	0.9374	9.0748	0.0625	0.0074	0.1045	0.9343
No log	14.96	180	0.1504	0.0625	0.9374	9.0912	0.0625	0.0074	0.1018	0.9358
No log	15.96	192	0.1504	0.0625	0.9374	9.0950	0.0625	0.0074	0.1032	0.9331
No log	16.96	204	0.1503	0.0625	0.9374	9.2141	0.0625	0.0074	0.1015	0.9363
No log	17.96	216	0.1503	0.0625	0.9374	9.0918	0.0625	0.0074	0.1046	0.9354
No log	18.96	228	0.1503	0.0625	0.9374	9.1430	0.0625	0.0074	0.1018	0.9385
No log	19.96	240	0.1503	0.0625	0.9374	9.2149	0.0625	0.0074	0.0991	0.9404
No log	20.96	252	0.1503	0.0625	0.9374	9.0900	0.0625	0.0074	0.1043	0.9386
No log	21.96	264	0.1503	0.0625	0.9374	9.1244	0.0625	0.0074	0.1060	0.9395
No log	22.96	276	0.1503	0.0625	0.9374	9.1353	0.0625	0.0074	0.1005	0.9378
No log	23.96	288	0.1502	0.0625	0.9374	9.2063	0.0625	0.0074	0.1032	0.9373
No log	24.96	300	0.1502	0.0625	0.9374	9.1398	0.0625	0.0074	0.1015	0.9383

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1.post200
• Datasets 2.9.0
• Tokenizers 0.13.2