SWIN_Gaudi_v1

This model is a fine-tuned version of microsoft/swinv2-large-patch4-window12-192-22k on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 4.8090
• Accuracy: 0.2293
• Memory Allocated (gb): 1.84
• Max Memory Allocated (gb): 58.8
• Total Memory Available (gb): 94.62

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: 0.001
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• total_train_batch_size: 1024
• total_eval_batch_size: 1024
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-06
• lr_scheduler_type: linear
• num_epochs: 30.0

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss	Memory Allocated (gb)	Allocated (gb)	Memory Available (gb)
5.3071	1.0	657	0.0801	6.2406	60.21	3.2	94.62
3.1366	2.0	1314	0.1047	5.8481	60.21	3.2	94.62
2.6048	3.0	1971	0.1238	5.5522	60.21	3.2	94.62
1.9918	4.0	2628	0.1301	5.5551	60.21	3.2	94.62
1.8353	5.0	3285	0.1415	5.4142	60.21	3.2	94.62
1.7262	6.0	3942	0.1495	5.4061	60.21	3.2	94.62
1.5135	7.0	4599	0.1468	5.4261	60.21	3.2	94.62
1.4225	8.0	5256	0.1573	5.3333	60.21	3.2	94.62
1.354	9.0	5913	0.1638	5.2205	60.21	3.2	94.62
1.2511	10.0	6570	0.1708	5.2129	60.21	3.2	94.62
1.1742	11.0	7227	0.1724	5.2002	60.21	3.2	94.62
1.1342	12.0	7884	0.1782	5.1635	60.21	3.2	94.62
1.0711	13.0	8541	0.1779	5.1436	60.21	3.2	94.62
0.9971	14.0	9198	0.1817	5.1076	60.21	3.2	94.62
0.9774	15.0	9855	0.1935	4.9076	60.21	3.2	94.62
0.9174	16.0	10512	0.1890	5.0318	60.21	3.2	94.62
0.8675	17.0	11169	0.1951	5.0392	60.21	3.2	94.62
0.8499	18.0	11826	0.1978	5.0243	60.21	3.2	94.62
0.8262	19.0	12483	0.1972	5.0843	60.21	3.2	94.62
0.7623	20.0	13140	0.2048	5.0004	60.21	3.2	94.62
0.7481	21.0	13797	0.2132	4.8428	60.24	3.2	94.62
0.7284	22.0	14454	0.2149	4.8461	60.24	3.2	94.62
0.6834	23.0	15111	0.2159	4.8741	60.24	3.2	94.62
0.6591	24.0	15768	0.2187	4.8993	60.24	3.2	94.62
0.6447	25.0	16425	0.2196	4.8415	60.24	3.2	94.62
0.6107	26.0	17082	0.2216	4.8600	60.24	3.2	94.62
0.5958	27.0	17739	0.2245	4.8391	60.24	3.2	94.62
0.5836	28.0	18396	0.2265	4.8561	60.24	3.2	94.62
0.5547	29.0	19053	0.2295	4.7933	60.24	3.2	94.62
0.547	30.0	19710	0.2293	4.8090	60.24	3.2	94.62

Framework versions

• Transformers 4.45.2
• Pytorch 2.6.0+hpu_1.20.0-543.git4952fce
• Datasets 3.5.0
• Tokenizers 0.20.3