swinv2-tiny-patch4-window8-256-ve-U13-b-80b

This model is a fine-tuned version of microsoft/swinv2-tiny-patch4-window8-256 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.7057
• Accuracy: 0.7391

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: 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: 80

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.92	6	1.3861	0.1304
1.386	2.0	13	1.3837	0.4348
1.386	2.92	19	1.3776	0.3043
1.3807	4.0	26	1.3570	0.2391
1.3386	4.92	32	1.3224	0.2174
1.3386	6.0	39	1.2085	0.3478
1.209	6.92	45	1.1056	0.4565
1.0561	8.0	52	1.0507	0.4783
1.0561	8.92	58	1.0161	0.4565
0.9157	10.0	65	0.8613	0.6304
0.8002	10.92	71	0.9073	0.5652
0.8002	12.0	78	0.8300	0.6304
0.7181	12.92	84	0.8958	0.5870
0.6405	14.0	91	0.8075	0.7174
0.6405	14.92	97	0.7478	0.6957
0.6064	16.0	104	0.7370	0.7174
0.5556	16.92	110	0.7057	0.7391
0.5556	18.0	117	0.7395	0.6522
0.4822	18.92	123	0.8734	0.6957
0.4241	20.0	130	0.9991	0.6739
0.4241	20.92	136	0.8416	0.7174
0.4307	22.0	143	0.9195	0.6957
0.4307	22.92	149	0.9211	0.6522
0.381	24.0	156	0.9683	0.6087
0.3707	24.92	162	1.0067	0.6739
0.3707	26.0	169	0.9793	0.6522
0.3918	26.92	175	0.9758	0.6739
0.3513	28.0	182	0.9761	0.6739
0.3513	28.92	188	1.0745	0.6304
0.2739	30.0	195	1.0775	0.6739
0.2882	30.92	201	1.1521	0.6739
0.2882	32.0	208	1.2072	0.6522
0.2588	32.92	214	1.1374	0.6739
0.2498	34.0	221	1.2131	0.6522
0.2498	34.92	227	1.1309	0.7391
0.2584	36.0	234	1.2828	0.6957
0.2228	36.92	240	1.1381	0.6739
0.2228	38.0	247	1.2116	0.6522
0.2408	38.92	253	1.1962	0.6739
0.2042	40.0	260	1.2557	0.6739
0.2042	40.92	266	1.3511	0.6739
0.2141	42.0	273	1.3636	0.6304
0.2141	42.92	279	1.3084	0.6304
0.2135	44.0	286	1.3847	0.6087
0.191	44.92	292	1.2408	0.6957
0.191	46.0	299	1.1750	0.7174
0.1833	46.92	305	1.1804	0.6957
0.189	48.0	312	1.1867	0.7174
0.189	48.92	318	1.0623	0.7391
0.2196	50.0	325	1.2626	0.6957
0.1505	50.92	331	1.2745	0.6957
0.1505	52.0	338	1.3473	0.6957
0.1604	52.92	344	1.3535	0.6522
0.1377	54.0	351	1.3873	0.6522
0.1377	54.92	357	1.4287	0.6522
0.1752	56.0	364	1.3014	0.6957
0.1684	56.92	370	1.3564	0.6739
0.1684	58.0	377	1.4165	0.6957
0.1597	58.92	383	1.3624	0.6739
0.1393	60.0	390	1.3018	0.6957
0.1393	60.92	396	1.3197	0.6739
0.1347	62.0	403	1.3542	0.6739
0.1347	62.92	409	1.3460	0.6739
0.155	64.0	416	1.3998	0.6739
0.1198	64.92	422	1.3982	0.6739
0.1198	66.0	429	1.3989	0.6522
0.1318	66.92	435	1.4035	0.6522
0.1382	68.0	442	1.3626	0.6522
0.1382	68.92	448	1.3714	0.6522
0.1451	70.0	455	1.4174	0.6739
0.1203	70.92	461	1.4343	0.6739
0.1203	72.0	468	1.4045	0.6522
0.141	72.92	474	1.3904	0.6522
0.1516	73.85	480	1.3849	0.6522

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2+cu118
• Datasets 2.16.1
• Tokenizers 0.15.0