microsoft-resnet-50-cartoon-emotion-detection

This model is a fine-tuned version of microsoft/resnet-50 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.4801
• Accuracy: 0.8165
• Precision: 0.8182
• Recall: 0.8165
• F1: 0.8173

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: 0.00012
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• 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	Precision	Recall	F1
No log	0.97	8	1.3855	0.2294	0.2697	0.2294	0.2165
1.4222	1.97	16	1.3792	0.2569	0.2808	0.2569	0.2543
1.4183	2.97	24	1.3646	0.3853	0.4102	0.3853	0.3511
1.4097	3.97	32	1.3563	0.4128	0.5062	0.4128	0.3245
1.3944	4.97	40	1.3462	0.4037	0.3927	0.4037	0.2939
1.3944	5.97	48	1.3223	0.4037	0.5152	0.4037	0.2841
1.411	6.97	56	1.3040	0.4128	0.4404	0.4128	0.2985
1.346	7.97	64	1.2700	0.4954	0.4960	0.4954	0.4093
1.3031	8.97	72	1.2150	0.5596	0.5440	0.5596	0.4672
1.2371	9.97	80	1.1580	0.5963	0.5659	0.5963	0.5101
1.2371	10.97	88	1.0670	0.6055	0.7279	0.6055	0.5211
1.1736	11.97	96	0.9856	0.6606	0.5537	0.6606	0.5772
1.0457	12.97	104	0.8963	0.6697	0.7631	0.6697	0.5965
0.953	13.97	112	0.8547	0.6697	0.6885	0.6697	0.6081
0.8579	14.97	120	0.7849	0.7156	0.7396	0.7156	0.6643
0.8579	15.97	128	0.7564	0.7431	0.7372	0.7431	0.7119
0.8167	16.97	136	0.7133	0.7615	0.7507	0.7615	0.7211
0.7273	17.97	144	0.6888	0.7523	0.7379	0.7523	0.7202
0.6547	18.97	152	0.6592	0.7798	0.7773	0.7798	0.7577
0.5963	19.97	160	0.6136	0.7706	0.7642	0.7706	0.7551
0.5963	20.97	168	0.5723	0.7890	0.7802	0.7890	0.7787
0.551	21.97	176	0.5686	0.7890	0.7761	0.7890	0.7781
0.4929	22.97	184	0.5597	0.7706	0.7649	0.7706	0.7651
0.4309	23.97	192	0.5234	0.7890	0.7774	0.7890	0.7810
0.3945	24.97	200	0.5008	0.7890	0.7837	0.7890	0.7813
0.3945	25.97	208	0.5289	0.7523	0.7537	0.7523	0.7529
0.3704	26.97	216	0.4399	0.7982	0.7958	0.7982	0.7963
0.3267	27.97	224	0.4539	0.8073	0.7983	0.8073	0.8005
0.2966	28.97	232	0.4735	0.7798	0.7892	0.7798	0.7837
0.2645	29.97	240	0.4594	0.7706	0.7706	0.7706	0.7706
0.2645	30.97	248	0.4699	0.7523	0.7554	0.7523	0.7533
0.2527	31.97	256	0.4551	0.7890	0.7856	0.7890	0.7857
0.2202	32.97	264	0.4458	0.8165	0.8198	0.8165	0.8170
0.2006	33.97	272	0.4632	0.7798	0.7941	0.7798	0.7850
0.1589	34.97	280	0.4651	0.7890	0.7993	0.7890	0.7925
0.1589	35.97	288	0.4595	0.7798	0.7824	0.7798	0.7804
0.153	36.97	296	0.4584	0.7615	0.7691	0.7615	0.7633
0.1427	37.97	304	0.4608	0.7798	0.7830	0.7798	0.7796
0.113	38.97	312	0.4571	0.7890	0.7922	0.7890	0.7899
0.1146	39.97	320	0.5270	0.7615	0.7651	0.7615	0.7613
0.1146	40.97	328	0.4888	0.7706	0.7782	0.7706	0.7710
0.1275	41.97	336	0.4523	0.7890	0.7809	0.7890	0.7837
0.0959	42.97	344	0.4697	0.7798	0.7753	0.7798	0.7767
0.0882	43.97	352	0.4286	0.7706	0.7686	0.7706	0.7686
0.0847	44.97	360	0.5317	0.7890	0.7993	0.7890	0.7925
0.0847	45.97	368	0.5431	0.7615	0.7700	0.7615	0.7647
0.0813	46.97	376	0.4432	0.8257	0.8435	0.8257	0.8284
0.0768	47.97	384	0.4886	0.7982	0.8005	0.7982	0.7956
0.0627	48.97	392	0.5373	0.7982	0.8072	0.7982	0.8010
0.0688	49.97	400	0.5897	0.7798	0.7892	0.7798	0.7822
0.0688	50.97	408	0.5115	0.7982	0.8015	0.7982	0.7992
0.0676	51.97	416	0.4881	0.7982	0.7998	0.7982	0.7978
0.0539	52.97	424	0.4820	0.8073	0.8139	0.8073	0.8077
0.0596	53.97	432	0.4450	0.8257	0.8246	0.8257	0.8244
0.0611	54.97	440	0.5057	0.7890	0.8008	0.7890	0.7924
0.0611	55.97	448	0.4918	0.7982	0.8056	0.7982	0.8008
0.0643	56.97	456	0.5946	0.7523	0.7587	0.7523	0.7545
0.0605	57.97	464	0.4888	0.8073	0.8239	0.8073	0.8121
0.063	58.97	472	0.5917	0.7890	0.8051	0.7890	0.7937
0.0595	59.97	480	0.5117	0.7890	0.7904	0.7890	0.7894
0.0595	60.97	488	0.5497	0.7615	0.7692	0.7615	0.7635
0.0554	61.97	496	0.4742	0.8165	0.8101	0.8165	0.8126
0.0557	62.97	504	0.5369	0.7890	0.7886	0.7890	0.7886
0.0539	63.97	512	0.5440	0.7890	0.7922	0.7890	0.7899
0.048	64.97	520	0.5924	0.7890	0.7878	0.7890	0.7883
0.048	65.97	528	0.4863	0.8440	0.8440	0.8440	0.8440
0.045	66.97	536	0.5850	0.8073	0.8076	0.8073	0.8047
0.047	67.97	544	0.4939	0.8257	0.8212	0.8257	0.8227
0.0412	68.97	552	0.4850	0.7890	0.7912	0.7890	0.7900
0.0392	69.97	560	0.5066	0.8257	0.8265	0.8257	0.8258
0.0392	70.97	568	0.4965	0.8073	0.8053	0.8073	0.8058
0.0423	71.97	576	0.4717	0.8349	0.8376	0.8349	0.8351
0.0471	72.97	584	0.4845	0.8257	0.8378	0.8257	0.8296
0.0322	73.97	592	0.5188	0.7706	0.7689	0.7706	0.7693
0.042	74.97	600	0.5242	0.7706	0.7699	0.7706	0.7701
0.042	75.97	608	0.5945	0.7798	0.7824	0.7798	0.7804
0.0416	76.97	616	0.5432	0.7982	0.8038	0.7982	0.7993
0.0399	77.97	624	0.5381	0.7982	0.8072	0.7982	0.7994
0.0439	78.97	632	0.6181	0.7798	0.7878	0.7798	0.7827
0.0462	79.97	640	0.4801	0.8165	0.8182	0.8165	0.8173

Framework versions

• Transformers 4.25.1
• Pytorch 1.13.1+cu117
• Datasets 2.8.0
• Tokenizers 0.11.0