tsf-gs-rot-flip-wtoken-DRPT-r128-f150-8.8-h768-i3072-p32-b8-e60

This model is a fine-tuned version of facebook/timesformer-base-finetuned-k400 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.7740
• Accuracy: 0.8610

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 6480
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.1114	0.0168	109	1.0984	0.3797
1.1049	1.0168	218	1.1141	0.3262
1.1056	2.0168	327	1.1710	0.3262
1.1154	3.0168	436	1.1256	0.3369
1.1056	4.0168	545	1.0962	0.3529
1.1569	5.0168	654	1.1136	0.3262
1.036	6.0168	763	1.0213	0.4652
1.0506	7.0168	872	1.0488	0.4278
1.042	8.0168	981	0.9366	0.5561
0.8999	9.0168	1090	0.8098	0.6310
0.9432	10.0168	1199	0.9513	0.5936
0.7898	11.0168	1308	0.5836	0.8021
0.7952	12.0168	1417	0.5680	0.7647
0.6641	13.0168	1526	0.6147	0.7861
0.6901	14.0168	1635	0.5688	0.7754
0.4637	15.0168	1744	0.5834	0.7914
0.5898	16.0168	1853	0.6636	0.7326
0.7036	17.0168	1962	0.7142	0.7433
0.3946	18.0168	2071	0.4866	0.8342
0.5379	19.0168	2180	0.6641	0.7701
0.5869	20.0168	2289	0.4817	0.8289
0.4564	21.0168	2398	0.4909	0.8396
0.419	22.0168	2507	0.5006	0.8235
0.4989	23.0168	2616	0.5648	0.8182
0.2701	24.0168	2725	0.5963	0.8342
0.5191	25.0168	2834	0.5766	0.7914
0.5088	26.0168	2943	0.4679	0.8610
0.3828	27.0168	3052	0.5231	0.8503
0.4228	28.0168	3161	0.6142	0.8235
0.5544	29.0168	3270	0.6508	0.8289
0.3595	30.0168	3379	0.6572	0.7914
0.3117	31.0168	3488	0.5587	0.8342
0.3324	32.0168	3597	0.5021	0.8610
0.3282	33.0168	3706	0.7642	0.8235
0.427	34.0168	3815	0.5739	0.8663
0.152	35.0168	3924	0.6957	0.8610
0.426	36.0168	4033	0.6705	0.8342
0.2803	37.0168	4142	0.5854	0.8449
0.3198	38.0168	4251	0.5280	0.8449
0.4348	39.0168	4360	0.7755	0.8128
0.1915	40.0168	4469	0.6813	0.8503
0.0793	41.0168	4578	0.7260	0.8503
0.3902	42.0168	4687	0.6581	0.8663
0.3552	43.0168	4796	0.5732	0.8610
0.3091	44.0168	4905	0.7510	0.8396
0.1103	45.0168	5014	0.7604	0.8503
0.3362	46.0168	5123	0.7156	0.8556
0.1935	47.0168	5232	0.6882	0.8503
0.0889	48.0168	5341	0.7639	0.8663
0.2156	49.0168	5450	0.8250	0.8610
0.0949	50.0168	5559	0.8256	0.8556
0.1735	51.0168	5668	0.6839	0.8770
0.1612	52.0168	5777	0.9082	0.8663
0.0556	53.0168	5886	0.7659	0.8770
0.0997	54.0168	5995	0.8025	0.8824
0.1648	55.0168	6104	0.8449	0.8770
0.1443	56.0168	6213	0.7801	0.8770
0.0405	57.0168	6322	0.8647	0.8663
0.3323	58.0168	6431	0.8411	0.8503
0.1137	59.0076	6480	0.7740	0.8610

Framework versions

• Transformers 4.41.2
• Pytorch 1.13.0+cu117
• Datasets 2.20.0
• Tokenizers 0.19.1