MAE-CT-M1N0-M12_v8_split5

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: 0.7761
• Accuracy: 0.7403

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: 6200

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6772	0.0102	63	0.6890	0.5185
0.7281	1.0102	126	0.7085	0.5185
0.7105	2.0102	189	0.7149	0.5185
0.5807	3.0102	252	0.7501	0.5185
0.4887	4.0102	315	0.9063	0.5185
0.7261	5.0102	378	0.7094	0.5185
0.6636	6.0102	441	0.6775	0.5185
0.6699	7.0102	504	0.6032	0.6296
0.3001	8.0102	567	1.2154	0.5556
0.2793	9.0102	630	1.1778	0.5556
0.5364	10.0102	693	1.2644	0.6296
0.2033	11.0102	756	0.6939	0.6667
0.4315	12.0102	819	1.1141	0.7037
0.4352	13.0102	882	1.3884	0.5556
0.6593	14.0102	945	1.6278	0.6296
0.6536	15.0102	1008	1.4868	0.6296
0.0819	16.0102	1071	2.1687	0.6667
0.0296	17.0102	1134	2.1161	0.5926
0.347	18.0102	1197	1.6357	0.6667
0.1476	19.0102	1260	2.8088	0.5556
0.0163	20.0102	1323	1.9632	0.5556
0.5644	21.0102	1386	2.1662	0.5185
0.1428	22.0102	1449	3.2763	0.5185
0.0068	23.0102	1512	3.2954	0.5185
0.0002	24.0102	1575	2.7854	0.5556
0.0048	25.0102	1638	3.1844	0.5185
0.0002	26.0102	1701	3.5731	0.4815
0.149	27.0102	1764	3.5418	0.5926
0.0007	28.0102	1827	3.6389	0.5185
0.0005	29.0102	1890	3.1077	0.5926
0.001	30.0102	1953	3.5467	0.5556
0.0003	31.0102	2016	3.7925	0.5556
0.0001	32.0102	2079	3.5266	0.5185
0.0009	33.0102	2142	3.4863	0.5926
0.0001	34.0102	2205	3.7751	0.5185
0.0001	35.0102	2268	3.5210	0.5185
0.2023	36.0102	2331	3.3674	0.5185
0.0001	37.0102	2394	3.9810	0.5185
0.0001	38.0102	2457	3.9022	0.5556
0.0001	39.0102	2520	3.9362	0.5556
0.0001	40.0102	2583	4.3247	0.5556
0.0001	41.0102	2646	4.3731	0.5185
0.0001	42.0102	2709	4.4895	0.5185
0.0002	43.0102	2772	3.7851	0.5556
0.0001	44.0102	2835	3.2867	0.5185
0.1972	45.0102	2898	4.1406	0.5556
0.0001	46.0102	2961	4.5674	0.4815
0.0001	47.0102	3024	3.8698	0.5556
0.0001	48.0102	3087	4.0334	0.4815
0.0001	49.0102	3150	4.1438	0.5185
0.1451	50.0102	3213	3.8312	0.5556
0.0001	51.0102	3276	2.6480	0.6667
0.0001	52.0102	3339	3.6914	0.4815
0.0012	53.0102	3402	3.7909	0.6296
0.0002	54.0102	3465	3.7010	0.5185
0.0001	55.0102	3528	3.7795	0.4815
0.0001	56.0102	3591	3.8197	0.5556
0.055	57.0102	3654	3.8996	0.5556
0.0001	58.0102	3717	4.1382	0.4815
0.0	59.0102	3780	4.3767	0.4444
0.0	60.0102	3843	4.4118	0.4444
0.0	61.0102	3906	4.5173	0.4444
0.0	62.0102	3969	4.5574	0.4444
0.0	63.0102	4032	4.5809	0.4444
0.0	64.0102	4095	4.6070	0.4444
0.0119	65.0102	4158	4.2245	0.5185
0.0001	66.0102	4221	4.0994	0.5185
0.0	67.0102	4284	4.0198	0.4815
0.0	68.0102	4347	4.0467	0.5185
0.0	69.0102	4410	4.0546	0.5185
0.0001	70.0102	4473	3.7226	0.5926
0.0001	71.0102	4536	3.6231	0.4815
0.0	72.0102	4599	3.5370	0.5926
0.0	73.0102	4662	3.9063	0.5185
0.0308	74.0102	4725	3.2158	0.5926
0.0001	75.0102	4788	3.7864	0.5556
0.0	76.0102	4851	3.7783	0.5556
0.0002	77.0102	4914	3.5353	0.6667
0.0	78.0102	4977	3.5016	0.6296
0.0	79.0102	5040	3.8000	0.5185
0.0	80.0102	5103	3.9970	0.5926
0.0	81.0102	5166	4.0152	0.5556
0.0	82.0102	5229	3.6760	0.6296
0.0	83.0102	5292	4.0728	0.5556
0.0	84.0102	5355	4.0964	0.5556
0.0	85.0102	5418	3.7875	0.5185
0.0	86.0102	5481	3.9301	0.5185
0.0	87.0102	5544	4.2208	0.5185
0.0	88.0102	5607	4.3575	0.5185
0.0	89.0102	5670	4.4015	0.5185
0.0	90.0102	5733	4.4241	0.5185
0.0	91.0102	5796	4.4360	0.5185
0.0	92.0102	5859	4.4440	0.5185
0.0	93.0102	5922	4.4219	0.5185
0.0	94.0102	5985	4.4249	0.5185
0.0	95.0102	6048	4.4252	0.5185
0.0	96.0102	6111	4.3802	0.5556
0.0	97.0102	6174	4.3659	0.5556
0.0	98.0042	6200	4.3659	0.5556

Framework versions

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