swinv2-small-panorama-IQA

This model is a fine-tuned version of microsoft/swinv2-small-patch4-window16-256 on the isiqa-2019-hf dataset. It achieves the following results on the evaluation set:

• Loss: 0.0223
• Srocc: 0.1291
• Lcc: 0.1271

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 32
• seed: 10
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 50.0

Training results

Training Loss	Epoch	Step	Validation Loss	Srocc	Lcc
No log	0.8571	3	0.2948	-0.3890	-0.3824
No log	2.0	7	0.1143	-0.3665	-0.3732
0.1552	2.8571	10	0.0768	-0.3477	-0.3657
0.1552	4.0	14	0.0748	-0.3395	-0.3504
0.1552	4.8571	17	0.0517	-0.3498	-0.3322
0.0657	6.0	21	0.0553	-0.3337	-0.3060
0.0657	6.8571	24	0.0434	-0.2921	-0.2810
0.0657	8.0	28	0.0406	-0.2481	-0.2570
0.0249	8.8571	31	0.0402	-0.2346	-0.2478
0.0249	10.0	35	0.0384	-0.2076	-0.2182
0.0249	10.8571	38	0.0317	-0.1919	-0.1923
0.0215	12.0	42	0.0310	-0.1518	-0.1636
0.0215	12.8571	45	0.0317	-0.1291	-0.1549
0.0215	14.0	49	0.0301	-0.0975	-0.1292
0.0154	14.8571	52	0.0285	-0.0804	-0.1057
0.0154	16.0	56	0.0277	-0.0461	-0.0762
0.0154	16.8571	59	0.0263	-0.0357	-0.0485
0.0128	18.0	63	0.0263	-0.0171	-0.0317
0.0128	18.8571	66	0.0265	-0.0040	-0.0236
0.0113	20.0	70	0.0263	0.0227	-0.0089
0.0113	20.8571	73	0.0256	0.0254	0.0081
0.0113	22.0	77	0.0249	0.0493	0.0233
0.0104	22.8571	80	0.0246	0.0616	0.0330
0.0104	24.0	84	0.0242	0.0691	0.0435
0.0104	24.8571	87	0.0240	0.0796	0.0518
0.0095	26.0	91	0.0238	0.0830	0.0679
0.0095	26.8571	94	0.0235	0.0929	0.0747
0.0095	28.0	98	0.0232	0.1003	0.0862
0.009	28.8571	101	0.0229	0.1050	0.0955
0.009	30.0	105	0.0226	0.1072	0.1052
0.009	30.8571	108	0.0226	0.1177	0.1110
0.0084	32.0	112	0.0225	0.1286	0.1152
0.0084	32.8571	115	0.0224	0.1296	0.1167
0.0084	34.0	119	0.0224	0.1296	0.1185
0.0085	34.8571	122	0.0224	0.1310	0.1200
0.0085	36.0	126	0.0224	0.1263	0.1221
0.0085	36.8571	129	0.0224	0.1249	0.1233
0.0082	38.0	133	0.0223	0.1272	0.1247
0.0082	38.8571	136	0.0223	0.1272	0.1255
0.008	40.0	140	0.0223	0.1291	0.1265
0.008	40.8571	143	0.0223	0.1291	0.1269
0.008	42.0	147	0.0223	0.1291	0.1271
0.0078	42.8571	150	0.0223	0.1291	0.1271

Framework versions

• Transformers 4.42.3
• Pytorch 2.1.2
• Datasets 2.20.0
• Tokenizers 0.19.1