hushem_1x_deit_tiny_adamax_lr0001_fold2

This model is a fine-tuned version of facebook/deit-tiny-patch16-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 1.9509
• Accuracy: 0.4667

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: 0.0001
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.67	1	1.8647	0.2444
No log	2.0	3	1.3951	0.2667
No log	2.67	4	1.3922	0.2889
No log	4.0	6	1.4419	0.2889
No log	4.67	7	1.4528	0.2889
No log	6.0	9	1.4637	0.3111
1.1875	6.67	10	1.3956	0.3333
1.1875	8.0	12	1.3768	0.4
1.1875	8.67	13	1.4359	0.3778
1.1875	10.0	15	1.4704	0.4
1.1875	10.67	16	1.4280	0.3778
1.1875	12.0	18	1.3838	0.4667
1.1875	12.67	19	1.4103	0.4444
0.4412	14.0	21	1.5312	0.4222
0.4412	14.67	22	1.6068	0.4444
0.4412	16.0	24	1.5834	0.4222
0.4412	16.67	25	1.5809	0.4222
0.4412	18.0	27	1.5887	0.4444
0.4412	18.67	28	1.6461	0.4222
0.0689	20.0	30	1.7954	0.4222
0.0689	20.67	31	1.8270	0.4444
0.0689	22.0	33	1.8461	0.4667
0.0689	22.67	34	1.8602	0.4667
0.0689	24.0	36	1.8842	0.4444
0.0689	24.67	37	1.8985	0.4444
0.0689	26.0	39	1.9148	0.4444
0.0084	26.67	40	1.9205	0.4222
0.0084	28.0	42	1.9310	0.4444
0.0084	28.67	43	1.9361	0.4444
0.0084	30.0	45	1.9427	0.4444
0.0084	30.67	46	1.9452	0.4667
0.0084	32.0	48	1.9490	0.4667
0.0084	32.67	49	1.9503	0.4667
0.0036	33.33	50	1.9509	0.4667

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1