tsf-newDS-DRPT-r224-f90-8.8-h768-i3072-p32-b8-e100

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: 1.4780
• Accuracy: 0.7478

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: 13100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.1939	0.01	131	1.1309	0.3673
1.1355	1.01	262	1.1368	0.3230
1.0944	2.01	393	1.1001	0.3628
1.151	3.01	524	1.1368	0.3540
1.1259	4.01	655	1.1324	0.3230
1.1406	5.01	786	1.0984	0.3540
1.1126	6.01	917	1.0994	0.3540
1.1169	7.01	1048	1.1456	0.3230
1.1217	8.01	1179	1.1333	0.3230
1.1227	9.01	1310	1.1024	0.3230
1.1136	10.01	1441	1.1115	0.3540
1.0942	11.01	1572	1.0910	0.3142
1.1089	12.01	1703	1.0973	0.3540
1.1148	13.01	1834	1.1086	0.3496
1.1019	14.01	1965	1.0919	0.3540
1.1264	15.01	2096	1.1035	0.3540
1.1235	16.01	2227	1.0961	0.3540
1.1438	17.01	2358	1.0864	0.3584
1.1092	18.01	2489	1.0938	0.3319
1.111	19.01	2620	1.1190	0.3496
1.0871	20.01	2751	1.1039	0.3584
1.0632	21.01	2882	1.1465	0.3673
1.0743	22.01	3013	1.1446	0.3628
1.0811	23.01	3144	1.1103	0.3584
1.1378	24.01	3275	1.1192	0.3628
1.0274	25.01	3406	1.1488	0.3230
1.0446	26.01	3537	1.1257	0.3407
1.1225	27.01	3668	1.1199	0.3363
1.0504	28.01	3799	1.1628	0.3496
1.1138	29.01	3930	1.1905	0.3319
1.066	30.01	4061	1.1344	0.3407
1.0567	31.01	4192	1.1359	0.4027
1.011	32.01	4323	1.1819	0.3628
1.0595	33.01	4454	1.1846	0.3761
1.028	34.01	4585	1.2150	0.3717
1.045	35.01	4716	1.1456	0.3496
1.0459	36.01	4847	1.0731	0.4646
1.0581	37.01	4978	1.2463	0.4292
0.9436	38.01	5109	1.1388	0.4425
0.9794	39.01	5240	1.1613	0.4513
0.8882	40.01	5371	1.1544	0.4381
1.0316	41.01	5502	1.0461	0.4779
0.8349	42.01	5633	1.0396	0.5088
0.8478	43.01	5764	1.0630	0.5442
0.8072	44.01	5895	1.1215	0.5177
0.7213	45.01	6026	1.1616	0.6018
0.7108	46.01	6157	1.1122	0.6106
0.6225	47.01	6288	1.1400	0.6106
0.5557	48.01	6419	0.9576	0.6283
0.4944	49.01	6550	1.3350	0.5487
0.7068	50.01	6681	0.9125	0.6504
0.5947	51.01	6812	2.0044	0.4956
0.645	52.01	6943	1.1295	0.5796
0.4251	53.01	7074	1.7297	0.5
0.573	54.01	7205	0.9968	0.6372
0.4283	55.01	7336	1.1135	0.6195
0.6225	56.01	7467	0.8792	0.7212
0.3876	57.01	7598	1.3363	0.6150
0.4729	58.01	7729	1.2033	0.6460
0.4922	59.01	7860	1.0137	0.6593
0.3925	60.01	7991	1.5002	0.6106
0.4234	61.01	8122	1.3914	0.6018
0.3847	62.01	8253	1.2090	0.6460
0.3739	63.01	8384	1.1537	0.6549
0.4808	64.01	8515	1.0365	0.7124
0.2926	65.01	8646	1.2063	0.6814
0.5116	66.01	8777	0.9150	0.7301
0.34	67.01	8908	1.1562	0.6903
0.452	68.01	9039	1.2344	0.6947
0.2936	69.01	9170	2.3964	0.5088
0.3911	70.01	9301	1.4071	0.6327
0.19	71.01	9432	1.3819	0.6991
0.3191	72.01	9563	1.7279	0.6460
0.2172	73.01	9694	1.2274	0.7257
0.2871	74.01	9825	1.4077	0.6947
0.3536	75.01	9956	1.2094	0.7301
0.2616	76.01	10087	1.7737	0.6372
0.3808	77.01	10218	1.7553	0.6549
0.3956	78.01	10349	1.3767	0.7035
0.2217	79.01	10480	1.2784	0.7035
0.3449	80.01	10611	1.0742	0.7611
0.3193	81.01	10742	1.1135	0.7566
0.3241	82.01	10873	1.3711	0.7345
0.1948	83.01	11004	1.1718	0.7389
0.4882	84.01	11135	1.1333	0.7655
0.3604	85.01	11266	1.1587	0.7566
0.3536	86.01	11397	1.4604	0.6947
0.3896	87.01	11528	1.7899	0.6770
0.2398	88.01	11659	1.3172	0.7566
0.252	89.01	11790	1.7039	0.6858
0.1858	90.01	11921	2.2136	0.6195
0.2268	91.01	12052	1.4825	0.6991
0.2984	92.01	12183	1.5829	0.6858
0.1323	93.01	12314	1.5580	0.6947
0.3251	94.01	12445	1.4773	0.7522
0.1103	95.01	12576	1.7728	0.6460
0.2054	96.01	12707	1.6074	0.6681
0.2131	97.01	12838	1.9007	0.6770
0.0364	98.01	12969	1.5574	0.6947
0.1295	99.01	13100	1.4780	0.7478

Framework versions

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