convnextv2-large-1k-224-finetuned-cassava-leaf-disease

This model is a fine-tuned version of facebook/convnextv2-large-1k-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.4210
• Accuracy: 0.8692

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: 5e-05
• train_batch_size: 240
• eval_batch_size: 240
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 960
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
8.2962	0.49	10	5.4110	0.0033
3.1666	0.99	20	2.0615	0.5883
1.4693	1.48	30	1.0935	0.6084
0.8718	1.98	40	0.7291	0.7463
0.6252	2.47	50	0.5894	0.7916
0.5198	2.96	60	0.5204	0.8299
0.4517	3.46	70	0.4658	0.8393
0.4266	3.95	80	0.4664	0.8407
0.4049	4.44	90	0.4337	0.8579
0.3817	4.94	100	0.4247	0.8523
0.3696	5.43	110	0.4146	0.8621
0.3577	5.93	120	0.4058	0.8607
0.3577	6.42	130	0.4047	0.8636
0.3354	6.91	140	0.3985	0.8617
0.3356	7.41	150	0.4025	0.8645
0.3286	7.9	160	0.4054	0.8673
0.3225	8.4	170	0.4062	0.8631
0.317	8.89	180	0.4007	0.8692
0.3101	9.38	190	0.3931	0.8701
0.293	9.88	200	0.3928	0.8682
0.2992	10.37	210	0.3942	0.8668
0.2968	10.86	220	0.3892	0.8692
0.2794	11.36	230	0.3988	0.8701
0.2707	11.85	240	0.3865	0.8762
0.2883	12.35	250	0.4040	0.8640
0.2784	12.84	260	0.3930	0.8692
0.2667	13.33	270	0.3985	0.8701
0.2642	13.83	280	0.4160	0.8668
0.2612	14.32	290	0.4086	0.8687
0.2586	14.81	300	0.3990	0.8668
0.2483	15.31	310	0.4111	0.8720
0.254	15.8	320	0.4082	0.8748
0.2283	16.3	330	0.4165	0.8668
0.246	16.79	340	0.4264	0.8692
0.2365	17.28	350	0.4185	0.8692
0.2388	17.78	360	0.4152	0.8650
0.2401	18.27	370	0.4169	0.8659
0.2334	18.77	380	0.4187	0.8696
0.2245	19.26	390	0.4192	0.8692
0.2291	19.75	400	0.4210	0.8692

Framework versions

• Transformers 4.37.2
• Pytorch 2.2.1
• Datasets 2.18.0
• Tokenizers 0.15.1