swin-transformer2

This model is a fine-tuned version of microsoft/swin-large-patch4-window12-384 on the None dataset. It achieves the following results on the evaluation set:

Loss: 2.2129
Accuracy: 0.6386
F1: 0.6328

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: 4
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
1.6336	0.9840	46	1.6510	0.2530	0.1876
1.2894	1.9893	93	1.2218	0.4458	0.3780
1.0959	2.9947	140	1.1383	0.5060	0.3518
1.0467	4.0	187	0.9372	0.5542	0.4352
0.9879	4.9840	233	1.0139	0.5301	0.4718
0.9086	5.9893	280	0.8822	0.6627	0.6359
0.9776	6.9947	327	1.0269	0.5542	0.5139
0.9715	8.0	374	0.7964	0.5663	0.5588
0.9049	8.9840	420	0.7839	0.5904	0.5346
0.8697	9.9893	467	1.0379	0.5663	0.4921
0.882	10.9947	514	0.9132	0.5663	0.5379
0.832	12.0	561	0.8513	0.5783	0.5008
0.7475	12.9840	607	0.7612	0.6627	0.6427
0.9056	13.9893	654	0.8431	0.6145	0.5725
0.9978	14.9947	701	0.7221	0.7108	0.6983
0.6956	16.0	748	0.7545	0.6145	0.5888
0.7185	16.9840	794	0.6561	0.6627	0.6499
0.8139	17.9893	841	0.7512	0.6506	0.6386
0.6837	18.9947	888	0.6491	0.6988	0.6849
0.5191	20.0	935	0.7290	0.6386	0.6336
0.6538	20.9840	981	0.8000	0.6988	0.6621
0.7912	21.9893	1028	1.0183	0.6145	0.5824
0.6093	22.9947	1075	0.9124	0.6506	0.6396
0.5312	24.0	1122	0.9098	0.6024	0.5581
0.6654	24.9840	1168	1.0432	0.5422	0.5028
0.5798	25.9893	1215	0.7369	0.6627	0.6553
0.506	26.9947	1262	0.9057	0.6265	0.6236
0.4638	28.0	1309	0.7950	0.6867	0.6644
0.371	28.9840	1355	1.0368	0.6627	0.6473
0.4721	29.9893	1402	0.8129	0.6747	0.6673
0.54	30.9947	1449	1.0379	0.6627	0.6491
0.3978	32.0	1496	1.3857	0.5904	0.5481
0.3503	32.9840	1542	1.0920	0.6024	0.5847
0.4407	33.9893	1589	1.1912	0.5904	0.5505
0.3786	34.9947	1636	1.5071	0.6024	0.5915
0.3482	36.0	1683	1.1161	0.6386	0.6240
0.2695	36.9840	1729	1.2040	0.5904	0.5704
0.2296	37.9893	1776	1.5781	0.5181	0.4691
0.2922	38.9947	1823	1.3713	0.6024	0.5879
0.1511	40.0	1870	1.1638	0.6506	0.6553
0.2814	40.9840	1916	1.3384	0.6988	0.6939
0.2196	41.9893	1963	1.2872	0.6506	0.6330
0.2477	42.9947	2010	1.5322	0.6627	0.6375
0.3296	44.0	2057	1.3479	0.6506	0.6353
0.2015	44.9840	2103	1.2521	0.6145	0.6044
0.3476	45.9893	2150	1.2464	0.6747	0.6641
0.189	46.9947	2197	1.4480	0.6506	0.6235
0.1852	48.0	2244	1.3611	0.6747	0.6594
0.2798	48.9840	2290	1.4427	0.6988	0.6957
0.1523	49.9893	2337	1.3352	0.6506	0.6450
0.1224	50.9947	2384	1.8088	0.6386	0.6201
0.0926	52.0	2431	1.4695	0.6506	0.6296
0.2071	52.9840	2477	1.4673	0.6867	0.6806
0.1063	53.9893	2524	1.4862	0.7108	0.6975
0.1831	54.9947	2571	1.4666	0.6506	0.6161
0.158	56.0	2618	1.8832	0.6988	0.6673
0.26	56.9840	2664	1.5855	0.6386	0.5986
0.1697	57.9893	2711	1.2184	0.7470	0.7434
0.2024	58.9947	2758	1.3524	0.6867	0.6682
0.2495	60.0	2805	1.7523	0.6627	0.6427
0.1247	60.9840	2851	1.7007	0.6506	0.6372
0.1436	61.9893	2898	1.9171	0.6386	0.6120
0.1438	62.9947	2945	1.8998	0.6265	0.5897
0.1137	64.0	2992	2.4028	0.5904	0.5498
0.1619	64.9840	3038	1.7087	0.7470	0.7473
0.1105	65.9893	3085	1.6545	0.6988	0.6975
0.1597	66.9947	3132	1.8024	0.6747	0.6758
0.0338	68.0	3179	1.8962	0.6747	0.6706
0.1184	68.9840	3225	2.1642	0.7108	0.7102
0.0878	69.9893	3272	2.0974	0.6506	0.6610
0.0963	70.9947	3319	1.8719	0.7108	0.7162
0.0827	72.0	3366	1.7538	0.6988	0.7000
0.0933	72.9840	3412	1.9357	0.6988	0.6988
0.0593	73.9893	3459	1.9924	0.6506	0.6420
0.0423	74.9947	3506	2.2029	0.6627	0.6702
0.0311	76.0	3553	1.9236	0.7108	0.7155
0.1881	76.9840	3599	1.9606	0.6747	0.6787
0.0566	77.9893	3646	2.1122	0.6265	0.6206
0.0266	78.9947	3693	2.1469	0.6506	0.6536
0.1015	80.0	3740	2.0335	0.6506	0.6587
0.1083	80.9840	3786	2.2123	0.6506	0.6509
0.0161	81.9893	3833	2.3094	0.6988	0.7064
0.0194	82.9947	3880	2.3315	0.6145	0.6101
0.113	84.0	3927	2.5276	0.6867	0.6908
0.0653	84.9840	3973	2.0321	0.6265	0.6263
0.0684	85.9893	4020	2.0302	0.6627	0.6706
0.1724	86.9947	4067	2.5865	0.5904	0.5860
0.028	88.0	4114	2.3814	0.5904	0.5804
0.0528	88.9840	4160	2.2804	0.6386	0.6410
0.0341	89.9893	4207	2.0635	0.5783	0.5736
0.0074	90.9947	4254	2.3491	0.6024	0.5993
0.0165	92.0	4301	2.2152	0.6145	0.6036
0.0157	92.9840	4347	2.3380	0.6145	0.6036
0.0544	93.9893	4394	2.3319	0.6265	0.6221
0.0577	94.9947	4441	2.2671	0.6265	0.6221
0.1516	96.0	4488	2.2034	0.6265	0.6204
0.0318	96.9840	4534	2.1932	0.6265	0.6204
0.043	97.9893	4581	2.2178	0.6265	0.6204
0.0099	98.3957	4600	2.2129	0.6386	0.6328

Framework versions

Transformers 4.45.2
Pytorch 2.4.0+cu121
Datasets 3.1.0
Tokenizers 0.20.1

masafresh/swin-transformer2

swin-transformer2

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

为此模型做种