vit-base-patch16-384-finetuned-humid-classes-3

This model is a fine-tuned version of google/vit-base-patch16-384 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.1595
• Accuracy: 0.9714
• F1 Macro: 0.8897
• Precision Macro: 0.9800
• Recall Macro: 0.8667
• Precision Dry: 0.95
• Recall Dry: 1.0
• F1 Dry: 0.9744
• Precision Firm: 1.0
• Recall Firm: 1.0
• F1 Firm: 1.0
• Precision Humid: 1.0
• Recall Humid: 1.0
• F1 Humid: 1.0
• Precision Lump: 0.95
• Recall Lump: 1.0
• F1 Lump: 0.9744
• Precision Rockies: 1.0
• Recall Rockies: 0.3333
• F1 Rockies: 0.5

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1 Macro	Precision Macro	Recall Macro	Precision Dry	Recall Dry	F1 Dry	Precision Firm	Recall Firm	F1 Firm	Precision Humid	Recall Humid	F1 Humid	Precision Lump	Recall Lump	F1 Lump	Precision Rockies	Recall Rockies	F1 Rockies
No log	1.0	6	1.3058	0.5429	0.3147	0.3646	0.3419	0.8333	0.2632	0.4	0.5897	0.92	0.7188	0.0	0.0	0.0	0.4	0.5263	0.4545	0.0	0.0	0.0
1.4465	2.0	12	0.8895	0.8571	0.5395	0.5119	0.5709	0.8636	1.0	0.9268	0.9231	0.96	0.9412	0.0	0.0	0.0	0.7727	0.8947	0.8293	0.0	0.0	0.0
1.4465	3.0	18	0.5465	0.8857	0.5579	0.5298	0.5895	0.9048	1.0	0.95	0.9259	1.0	0.9615	0.0	0.0	0.0	0.8182	0.9474	0.8780	0.0	0.0	0.0
0.7112	4.0	24	0.3443	0.9143	0.6994	0.7447	0.6895	0.9048	1.0	0.95	0.9615	1.0	0.9804	1.0	0.5	0.6667	0.8571	0.9474	0.9	0.0	0.0	0.0
0.3288	5.0	30	0.3331	0.8857	0.7962	0.8415	0.7741	0.9474	0.9474	0.9474	1.0	0.84	0.9130	1.0	0.75	0.8571	0.76	1.0	0.8636	0.5	0.3333	0.4
0.3288	6.0	36	0.1950	0.9571	0.8643	0.8795	0.8561	0.9474	0.9474	0.9474	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	0.5	0.3333	0.4
0.1812	7.0	42	0.1595	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.1812	8.0	48	0.1558	0.9429	0.8155	0.8323	0.8061	0.95	1.0	0.9744	0.9615	1.0	0.9804	0.75	0.75	0.75	1.0	0.9474	0.9730	0.5	0.3333	0.4
0.0962	9.0	54	0.1787	0.9286	0.8127	0.7982	0.8351	0.9412	0.8421	0.8889	1.0	1.0	1.0	0.8	1.0	0.8889	1.0	1.0	1.0	0.25	0.3333	0.2857
0.1039	10.0	60	0.2412	0.9286	0.7421	0.7533	0.7395	0.9048	1.0	0.95	0.9615	1.0	0.9804	1.0	0.75	0.8571	0.9	0.9474	0.9231	0.0	0.0	0.0
0.1039	11.0	66	0.2776	0.9286	0.7670	0.7537	0.784	0.8636	1.0	0.9268	1.0	0.92	0.9583	1.0	1.0	1.0	0.9048	1.0	0.95	0.0	0.0	0.0
0.0457	12.0	72	0.1497	0.9429	0.8379	0.8418	0.8456	0.9474	0.9474	0.9474	0.9615	1.0	0.9804	0.8	1.0	0.8889	1.0	0.9474	0.9730	0.5	0.3333	0.4
0.0457	13.0	78	0.2168	0.9571	0.7849	0.7710	0.8	0.9048	1.0	0.95	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	0.0	0.0	0.0
0.0474	14.0	84	0.1866	0.9429	0.7762	0.7627	0.792	0.8636	1.0	0.9268	1.0	0.96	0.9796	1.0	1.0	1.0	0.95	1.0	0.9744	0.0	0.0	0.0
0.0165	15.0	90	0.1930	0.9571	0.8808	0.9710	0.8587	0.95	1.0	0.9744	1.0	0.96	0.9796	1.0	1.0	1.0	0.9048	1.0	0.95	1.0	0.3333	0.5
0.0165	16.0	96	0.2238	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.01	17.0	102	0.2782	0.9571	0.7849	0.7710	0.8	0.9048	1.0	0.95	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	0.0	0.0	0.0
0.01	18.0	108	0.2191	0.9429	0.8554	0.8704	0.8481	0.9474	0.9474	0.9474	1.0	0.96	0.9796	1.0	1.0	1.0	0.9048	1.0	0.95	0.5	0.3333	0.4
0.008	19.0	114	0.2492	0.9143	0.8037	0.8302	0.7851	0.9474	0.9474	0.9474	0.9259	1.0	0.9615	1.0	0.75	0.8571	0.9444	0.8947	0.9189	0.3333	0.3333	0.3333
0.0116	20.0	120	0.3098	0.9286	0.7427	0.7537	0.742	0.9048	1.0	0.95	1.0	0.96	0.9796	1.0	0.75	0.8571	0.8636	1.0	0.9268	0.0	0.0	0.0
0.0116	21.0	126	0.2063	0.9429	0.8710	0.9641	0.8456	0.95	1.0	0.9744	0.9259	1.0	0.9615	1.0	1.0	1.0	0.9444	0.8947	0.9189	1.0	0.3333	0.5
0.0069	22.0	132	0.1729	0.9429	0.8550	0.8713	0.8456	0.9474	0.9474	0.9474	0.9615	1.0	0.9804	1.0	1.0	1.0	0.9474	0.9474	0.9474	0.5	0.3333	0.4
0.0069	23.0	138	0.2078	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0067	24.0	144	0.2262	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0067	25.0	150	0.2329	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0067	26.0	156	0.2344	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0062	27.0	162	0.2380	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0062	28.0	168	0.2367	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0027	29.0	174	0.2372	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0057	30.0	180	0.2371	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0057	31.0	186	0.2362	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.004	32.0	192	0.2391	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.004	33.0	198	0.2348	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0043	34.0	204	0.2338	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0041	35.0	210	0.2370	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0041	36.0	216	0.2410	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0053	37.0	222	0.2430	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0053	38.0	228	0.2448	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0019	39.0	234	0.2432	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0051	40.0	240	0.2421	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0051	41.0	246	0.2451	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0035	42.0	252	0.2440	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0035	43.0	258	0.2472	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0065	44.0	264	0.2468	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0021	45.0	270	0.2439	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0021	46.0	276	0.2411	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.004	47.0	282	0.2368	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.004	48.0	288	0.2334	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0041	49.0	294	0.2311	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5
0.0032	50.0	300	0.2305	0.9714	0.8897	0.9800	0.8667	0.95	1.0	0.9744	1.0	1.0	1.0	1.0	1.0	1.0	0.95	1.0	0.9744	1.0	0.3333	0.5

Framework versions

• Transformers 4.56.1
• Pytorch 2.5.1+cu124
• Datasets 4.0.0
• Tokenizers 0.22.0