MAE-CT-M1N0-M12_v8_split4

This model is a fine-tuned version of MCG-NJU/videomae-large-finetuned-kinetics on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.4312
• Accuracy: 0.8267

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: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 6400

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6773	0.0102	65	0.7109	0.4348
0.7393	1.0102	130	0.7720	0.4348
0.6483	2.0102	195	0.8131	0.4348
0.5872	3.0102	260	0.7178	0.4348
0.5612	4.0102	325	0.6203	0.6957
0.2855	5.0102	390	0.7647	0.3913
0.3332	6.0102	455	0.9563	0.3913
0.5376	7.0102	520	1.0380	0.4348
0.3236	8.0102	585	0.6013	0.7826
0.2583	9.0102	650	0.6642	0.6957
0.519	10.0102	715	0.8797	0.6522
0.2594	11.0102	780	0.8123	0.7391
0.2015	12.0102	845	1.2630	0.6522
0.3333	13.0102	910	1.4962	0.6087
0.1593	14.0102	975	1.1972	0.6957
0.1296	15.0102	1040	1.1893	0.7826
0.3097	16.0102	1105	1.5245	0.7391
0.1145	17.0102	1170	1.2979	0.7826
0.2288	18.0102	1235	1.7658	0.6957
0.0217	19.0102	1300	2.6377	0.6087
0.1368	20.0102	1365	1.6947	0.6957
0.1717	21.0102	1430	1.8905	0.6522
0.0014	22.0102	1495	2.1503	0.6522
0.012	23.0102	1560	2.0506	0.6522
0.0007	24.0102	1625	2.3373	0.6522
0.0001	25.0102	1690	1.6162	0.7391
0.0002	26.0102	1755	2.7662	0.6087
0.104	27.0102	1820	1.5637	0.7826
0.1848	28.0102	1885	3.6887	0.5217
0.0015	29.0102	1950	1.7133	0.6957
0.0001	30.0102	2015	2.1864	0.7391
0.0008	31.0102	2080	1.9452	0.7391
0.0002	32.0102	2145	1.7982	0.7391
0.0001	33.0102	2210	2.3272	0.6957
0.0072	34.0102	2275	2.5865	0.6957
0.275	35.0102	2340	4.0065	0.5652
0.0004	36.0102	2405	1.4350	0.7826
0.0001	37.0102	2470	1.8396	0.7826
0.1562	38.0102	2535	2.6788	0.6522
0.0001	39.0102	2600	2.0010	0.6957
0.0001	40.0102	2665	2.4220	0.6522
0.1117	41.0102	2730	2.3290	0.6957
0.0001	42.0102	2795	3.1235	0.5652
0.0001	43.0102	2860	2.9064	0.6087
0.0003	44.0102	2925	3.1359	0.6087
0.0007	45.0102	2990	3.1225	0.6087
0.0031	46.0102	3055	2.9252	0.6087
0.0	47.0102	3120	3.3919	0.5652
0.0003	48.0102	3185	2.8240	0.6957
0.0014	49.0102	3250	2.4431	0.5652
0.0001	50.0102	3315	2.2488	0.6957
0.0	51.0102	3380	2.6169	0.6087
0.0	52.0102	3445	2.4118	0.7391
0.0002	53.0102	3510	2.4928	0.5652
0.0001	54.0102	3575	3.6149	0.5652
0.0	55.0102	3640	3.2978	0.5652
0.0	56.0102	3705	2.9060	0.5217
0.1108	57.0102	3770	3.0361	0.6087
0.0	58.0102	3835	3.3929	0.6087
0.0	59.0102	3900	3.5174	0.5652
0.0007	60.0102	3965	2.1117	0.7391
0.0	61.0102	4030	3.5274	0.6087
0.0	62.0102	4095	3.5149	0.6087
0.0	63.0102	4160	3.4865	0.6087
0.0	64.0102	4225	3.2318	0.6087
0.0	65.0102	4290	3.1844	0.6087
0.0	66.0102	4355	3.2181	0.6087
0.0	67.0102	4420	3.2936	0.6087
0.0	68.0102	4485	3.3043	0.6087
0.0	69.0102	4550	3.1360	0.6522
0.0186	70.0102	4615	2.3659	0.7391
0.0	71.0102	4680	2.5226	0.7391
0.0	72.0102	4745	2.7737	0.6522
0.0	73.0102	4810	2.6730	0.6957
0.0	74.0102	4875	2.7865	0.6957
0.0	75.0102	4940	2.7922	0.6957
0.0	76.0102	5005	3.0552	0.6087
0.0	77.0102	5070	2.4933	0.7391
0.0044	78.0102	5135	2.1811	0.7391
0.0	79.0102	5200	1.9051	0.7826
0.0	80.0102	5265	1.8407	0.8261
0.0	81.0102	5330	2.1967	0.7826
0.0	82.0102	5395	2.3231	0.6957
0.0	83.0102	5460	2.3425	0.6957
0.0	84.0102	5525	2.8403	0.5652
0.0	85.0102	5590	2.3424	0.6957
0.0	86.0102	5655	2.4246	0.6957
0.0	87.0102	5720	2.4289	0.6957
0.0	88.0102	5785	2.4310	0.6957
0.0	89.0102	5850	2.4361	0.6957
0.0	90.0102	5915	2.3667	0.6957
0.0	91.0102	5980	2.3627	0.6957
0.0	92.0102	6045	2.3715	0.6957
0.0	93.0102	6110	2.3773	0.7391
0.0	94.0102	6175	2.4264	0.7391
0.0	95.0102	6240	2.4393	0.7391
0.0	96.0102	6305	2.4449	0.7391
0.0	97.0102	6370	2.4451	0.7391
0.0	98.0047	6400	2.4451	0.7391

Framework versions

• Transformers 4.46.2
• Pytorch 2.0.1+cu117
• Datasets 3.0.1
• Tokenizers 0.20.0