cards_bottom_left_swin-tiny-patch4-window7-224-finetuned-dough_100_epochs

This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 1.0025
• Accuracy: 0.5947

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: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.6956	1.0	1252	1.4843	0.3970
1.5633	2.0	2504	1.2584	0.4782
1.5568	3.0	3756	1.1976	0.4918
1.4727	4.0	5009	1.1884	0.4916
1.468	5.0	6261	1.1909	0.4889
1.4663	6.0	7513	1.1263	0.5288
1.4409	7.0	8765	1.0967	0.5441
1.4329	8.0	10018	1.0976	0.5388
1.4842	9.0	11270	1.1076	0.5315
1.4253	10.0	12522	1.0634	0.5511
1.3888	11.0	13774	1.0489	0.5634
1.3681	12.0	15027	1.0663	0.5567
1.3802	13.0	16279	1.0304	0.5667
1.4016	14.0	17531	1.0592	0.5518
1.376	15.0	18783	1.0080	0.5776
1.3539	16.0	20036	1.0103	0.5742
1.3725	17.0	21288	1.0261	0.5636
1.3104	18.0	22540	1.0304	0.5686
1.3448	19.0	23792	1.0184	0.5687
1.3479	20.0	25045	0.9968	0.5809
1.3517	21.0	26297	1.1350	0.5182
1.3367	22.0	27549	0.9835	0.5867
1.3002	23.0	28801	1.0193	0.5736
1.3238	24.0	30054	0.9820	0.5875
1.2865	25.0	31306	1.0267	0.5617
1.3029	26.0	32558	1.0086	0.5730
1.3173	27.0	33810	0.9750	0.5924
1.297	28.0	35063	0.9851	0.5848
1.3105	29.0	36315	1.0306	0.5685
1.3477	30.0	37567	0.9977	0.5845
1.2565	31.0	38819	0.9900	0.5851
1.2657	32.0	40072	1.0137	0.5862
1.2911	33.0	41324	0.9947	0.5889
1.2539	34.0	42576	0.9821	0.5914
1.2441	35.0	43828	1.0296	0.5763
1.2176	36.0	45081	1.0350	0.5806
1.25	37.0	46333	1.0195	0.5779
1.2647	38.0	47585	1.0021	0.5903
1.2428	39.0	48837	1.0087	0.5892
1.2364	40.0	50090	1.0025	0.5947
1.2083	41.0	51342	1.0427	0.5862
1.2002	42.0	52594	1.0303	0.5878
1.2071	43.0	53846	1.0190	0.5909
1.1536	44.0	55099	1.0314	0.5920
1.2029	45.0	56351	1.0570	0.5839
1.2249	46.0	57603	1.0508	0.5828
1.1913	47.0	58855	1.0493	0.5853
1.1938	48.0	60108	1.0575	0.5857
1.1724	49.0	61360	1.0700	0.5905
1.1536	50.0	62612	1.0841	0.5853
1.1239	51.0	63864	1.0803	0.5865
1.1743	52.0	65117	1.0864	0.5880
1.1414	53.0	66369	1.1224	0.5819
1.1411	54.0	67621	1.1316	0.5780
1.1029	55.0	68873	1.1070	0.5860
1.1353	56.0	70126	1.1247	0.5847
1.1293	57.0	71378	1.1279	0.5805
1.1335	58.0	72630	1.1482	0.5812
1.1157	59.0	73882	1.1960	0.5674
1.0891	60.0	75135	1.1414	0.5848
1.1299	61.0	76387	1.1658	0.5790
1.0828	62.0	77639	1.1753	0.5806
1.0866	63.0	78891	1.1767	0.5755
1.0721	64.0	80144	1.1861	0.5808
1.0682	65.0	81396	1.2083	0.5749
1.0747	66.0	82648	1.2204	0.5755
1.0902	67.0	83900	1.2175	0.5750
1.0381	68.0	85153	1.2445	0.5738
1.049	69.0	86405	1.2674	0.5707
1.0501	70.0	87657	1.2602	0.5740
1.0117	71.0	88909	1.2549	0.5687
1.0179	72.0	90162	1.3010	0.5690
1.0788	73.0	91414	1.2723	0.5726
1.0234	74.0	92666	1.3162	0.5717
1.0325	75.0	93918	1.3136	0.5692
1.0079	76.0	95171	1.3337	0.5655
1.058	77.0	96423	1.3171	0.5719
0.9968	78.0	97675	1.3470	0.5693
1.0217	79.0	98927	1.3418	0.5733
1.0124	80.0	100180	1.3518	0.5700
0.9823	81.0	101432	1.3646	0.5700
0.9627	82.0	102684	1.3658	0.5686
0.9773	83.0	103936	1.3811	0.5674
0.9855	84.0	105189	1.4082	0.5638
0.9928	85.0	106441	1.3877	0.5612
1.0025	86.0	107693	1.3925	0.5653
0.9583	87.0	108945	1.4313	0.5625
0.977	88.0	110198	1.4153	0.5651
0.9825	89.0	111450	1.4426	0.5619
0.9315	90.0	112702	1.4376	0.5643
0.8916	91.0	113954	1.4630	0.5618
0.9495	92.0	115207	1.4501	0.5627
0.9372	93.0	116459	1.4606	0.5622
0.9284	94.0	117711	1.4725	0.5608
0.9266	95.0	118963	1.4680	0.5607
0.8858	96.0	120216	1.4705	0.5626
0.9025	97.0	121468	1.4818	0.5616
0.902	98.0	122720	1.4871	0.5606
0.8961	99.0	123972	1.4881	0.5612
0.9204	99.98	125200	1.4894	0.5609

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.1+cu117
• Datasets 2.17.0
• Tokenizers 0.13.3