tsf-newDS-r224-f90-8.8-h768-i3072-p32-b8-e50

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: 0.9773
• Accuracy: 0.8186

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: 6550
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.1214	0.02	131	1.1007	0.3319
1.135	1.02	262	1.1006	0.3496
1.1075	2.02	393	1.1293	0.3230
1.157	3.02	524	1.1034	0.3540
1.1213	4.02	655	1.2115	0.3230
1.1139	5.02	786	1.0907	0.3540
1.1237	6.02	917	1.0917	0.3230
1.1137	7.02	1048	1.1317	0.3230
1.1056	8.02	1179	1.1004	0.3319
1.1082	9.02	1310	1.1200	0.3142
1.0915	10.02	1441	1.0995	0.3982
1.0066	11.02	1572	1.0283	0.4602
0.9384	12.02	1703	1.0246	0.4690
0.991	13.02	1834	1.0044	0.4513
0.9787	14.02	1965	0.9504	0.5442
0.8522	15.02	2096	0.8761	0.6195
0.9404	16.02	2227	0.9207	0.5531
0.8297	17.02	2358	0.9401	0.6018
0.6493	18.02	2489	0.8470	0.6283
0.6117	19.02	2620	0.9770	0.6283
0.676	20.02	2751	0.9361	0.6504
0.5106	21.02	2882	0.7691	0.6947
0.5782	22.02	3013	0.6593	0.7832
0.5183	23.02	3144	0.9417	0.6770
0.4466	24.02	3275	0.7373	0.7257
0.7414	25.02	3406	0.7389	0.7788
0.6338	26.02	3537	0.6562	0.7788
0.5423	27.02	3668	0.4732	0.8540
0.6629	28.02	3799	0.9399	0.7434
0.3087	29.02	3930	1.0425	0.6947
0.2584	30.02	4061	0.6969	0.7965
0.2846	31.02	4192	0.6573	0.8186
0.4084	32.02	4323	0.6418	0.8053
0.4092	33.02	4454	1.1458	0.7434
0.4956	34.02	4585	0.8430	0.8053
0.28	35.02	4716	0.7289	0.8009
0.2646	36.02	4847	0.6761	0.8053
0.313	37.02	4978	1.0661	0.7832
0.2937	38.02	5109	0.9224	0.8009
0.2626	39.02	5240	0.5193	0.8628
0.4277	40.02	5371	0.6577	0.8319
0.1663	41.02	5502	1.1011	0.7743
0.1576	42.02	5633	0.7574	0.8407
0.3203	43.02	5764	0.8206	0.8407
0.2027	44.02	5895	1.1296	0.7743
0.188	45.02	6026	0.9082	0.8009
0.1701	46.02	6157	1.0158	0.8009
0.2083	47.02	6288	1.1480	0.7788
0.1618	48.02	6419	0.9984	0.8009
0.2421	49.02	6550	0.9773	0.8186

Framework versions

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