MAE-CT-CPC-Dicotomized-v7-tricot

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: 5.4223
• Accuracy: 0.3077

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 7900

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.0989	0.0101	80	1.1031	0.3191
1.0889	1.0101	160	1.1058	0.3404
1.0739	2.0101	240	1.1233	0.4043
1.0036	3.0101	320	1.1596	0.2766
1.0706	4.0101	400	1.1731	0.2553
0.9669	5.0101	480	1.1228	0.3191
1.0233	6.0101	560	1.1490	0.4043
0.8492	7.0101	640	1.2636	0.3830
0.8842	8.0101	720	1.4061	0.3617
0.6599	9.0101	800	1.3445	0.2979
0.6723	10.0101	880	1.4072	0.3617
0.604	11.0101	960	1.4199	0.3617
0.4959	12.0101	1040	1.5689	0.3617
0.3758	13.0101	1120	1.7867	0.3617
0.6257	14.0101	1200	1.9218	0.3617
0.3693	15.0101	1280	2.0988	0.3191
0.5933	16.0101	1360	1.8413	0.4043
0.202	17.0101	1440	2.7537	0.3191
0.1454	18.0101	1520	2.4612	0.4255
0.1332	19.0101	1600	3.0944	0.3404
0.9193	20.0101	1680	2.8691	0.4043
0.1201	21.0101	1760	3.0564	0.4255
0.1716	22.0101	1840	3.3907	0.3404
0.0402	23.0101	1920	3.7917	0.3191
0.0709	24.0101	2000	3.5487	0.4043
0.1021	25.0101	2080	3.9004	0.4043
0.0029	26.0101	2160	4.1949	0.3617
0.1352	27.0101	2240	4.5038	0.3617
0.0173	28.0101	2320	3.9352	0.3830
0.0012	29.0101	2400	4.3234	0.4043
0.0007	30.0101	2480	4.2877	0.3830
0.2292	31.0101	2560	4.7297	0.3191
0.0004	32.0101	2640	4.4710	0.3830
0.0361	33.0101	2720	4.2391	0.4255
0.0002	34.0101	2800	4.2256	0.4043
0.0082	35.0101	2880	5.0734	0.3404
0.0318	36.0101	2960	4.0735	0.4255
0.0002	37.0101	3040	5.1464	0.2553
0.0003	38.0101	3120	4.6340	0.4043
0.48	39.0101	3200	4.3370	0.4255
0.0002	40.0101	3280	4.5820	0.3617
0.0002	41.0101	3360	5.0157	0.3191
0.1209	42.0101	3440	4.3109	0.3830
0.0001	43.0101	3520	4.4596	0.4043
0.0109	44.0101	3600	4.4251	0.3830
0.0001	45.0101	3680	5.2962	0.2979
0.1516	46.0101	3760	4.2314	0.4043
0.0001	47.0101	3840	4.0705	0.5319
0.0001	48.0101	3920	4.5586	0.4255
0.0266	49.0101	4000	4.9479	0.4043
0.0001	50.0101	4080	4.3270	0.4468
0.1307	51.0101	4160	4.7948	0.3830
0.0019	52.0101	4240	4.3638	0.3617
0.0001	53.0101	4320	4.5863	0.4255
0.0001	54.0101	4400	4.7373	0.4255
0.0006	55.0101	4480	3.9066	0.4468
0.0001	56.0101	4560	4.0314	0.4681
0.0001	57.0101	4640	4.0581	0.5106
0.0001	58.0101	4720	5.0045	0.3830
0.0001	59.0101	4800	4.0895	0.4255
0.0713	60.0101	4880	5.0429	0.4255
0.0017	61.0101	4960	4.7870	0.4255
0.0676	62.0101	5040	5.0957	0.3830
0.0	63.0101	5120	4.6062	0.4043
0.0045	64.0101	5200	5.2459	0.3830
0.0943	65.0101	5280	5.0856	0.3617
0.0002	66.0101	5360	4.4492	0.4894
0.0002	67.0101	5440	5.1795	0.4043
0.0007	68.0101	5520	4.3202	0.4681
0.1678	69.0101	5600	4.8688	0.4043
0.0001	70.0101	5680	5.2880	0.4043
0.0	71.0101	5760	5.1151	0.4255
0.0005	72.0101	5840	4.5667	0.4255
0.0	73.0101	5920	4.2883	0.4681
0.0	74.0101	6000	4.6848	0.4255
0.0	75.0101	6080	4.8157	0.4468
0.0	76.0101	6160	4.8248	0.4468
0.0	77.0101	6240	4.5636	0.4894
0.0	78.0101	6320	4.5817	0.4255
0.0001	79.0101	6400	4.7743	0.3830
0.0001	80.0101	6480	4.9000	0.4043
0.0002	81.0101	6560	4.7669	0.4255
0.0	82.0101	6640	4.8225	0.4468
0.0	83.0101	6720	4.8331	0.4468
0.0	84.0101	6800	4.7154	0.4468
0.0	85.0101	6880	4.7169	0.4468
0.0	86.0101	6960	4.9004	0.4255
0.0	87.0101	7040	4.9092	0.4255
0.0	88.0101	7120	4.8941	0.4255
0.0	89.0101	7200	4.7898	0.4255
0.0	90.0101	7280	4.8271	0.4468
0.0	91.0101	7360	4.8320	0.4468
0.0	92.0101	7440	4.8274	0.4468
0.0	93.0101	7520	4.8269	0.4468
0.0	94.0101	7600	4.8785	0.3830
0.0	95.0101	7680	4.9640	0.4255
0.0	96.0101	7760	4.9480	0.4255
0.0	97.0101	7840	4.9404	0.4255
0.0	98.0076	7900	4.9351	0.4255

Framework versions

• Transformers 4.45.2
• Pytorch 2.0.1+cu117
• Datasets 3.0.1
• Tokenizers 0.20.0