MAE-CT-M1N0-M12_v8_split2

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: 0.5597
• Accuracy: 0.7260

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
0.6375	0.0102	63	0.6993	0.5161
0.7565	1.0102	126	0.7252	0.5161
0.6926	2.0102	189	0.7296	0.5161
0.5636	3.0102	252	0.7618	0.5161
0.4721	4.0102	315	1.2407	0.5161
0.7569	5.0102	378	0.8010	0.5161
0.384	6.0102	441	0.6036	0.6774
0.6542	7.0102	504	0.5455	0.8065
0.3615	8.0102	567	1.3506	0.5484
0.2246	9.0102	630	1.5499	0.5806
0.7929	10.0102	693	1.0719	0.6452
0.5963	11.0102	756	0.9215	0.6129
0.1342	12.0102	819	0.9188	0.6452
0.2511	13.0102	882	1.4410	0.6452
0.5877	14.0102	945	2.3550	0.5161
0.3261	15.0102	1008	1.0729	0.6774
0.0425	16.0102	1071	2.5330	0.5806
0.174	17.0102	1134	2.8190	0.5806
0.1972	18.0102	1197	2.4491	0.5484
0.2264	19.0102	1260	1.9345	0.6774
0.0862	20.0102	1323	3.3695	0.5161
0.0998	21.0102	1386	1.4091	0.7742
0.311	22.0102	1449	2.7629	0.5484
0.0481	23.0102	1512	2.0506	0.6452
0.2109	24.0102	1575	2.5990	0.5806
0.179	25.0102	1638	2.7815	0.5806
0.0002	26.0102	1701	3.6719	0.5161
0.0996	27.0102	1764	3.7618	0.5161
0.0002	28.0102	1827	3.3375	0.5484
0.0004	29.0102	1890	2.8750	0.6129
0.0001	30.0102	1953	2.5867	0.6774
0.1188	31.0102	2016	1.8263	0.6774
0.0295	32.0102	2079	3.2699	0.5806
0.1931	33.0102	2142	3.3532	0.5806
0.0002	34.0102	2205	4.2001	0.5161
0.0001	35.0102	2268	3.3819	0.5484
0.0001	36.0102	2331	2.2776	0.7097
0.0007	37.0102	2394	2.8516	0.5806
0.0001	38.0102	2457	4.0420	0.5161
0.0002	39.0102	2520	2.5901	0.6452
0.0001	40.0102	2583	3.5043	0.5806
0.0001	41.0102	2646	3.5424	0.5806
0.0001	42.0102	2709	3.8740	0.5484
0.0001	43.0102	2772	3.5726	0.5806
0.0004	44.0102	2835	3.2184	0.5806
0.0	45.0102	2898	3.3347	0.5806
0.0001	46.0102	2961	3.8206	0.5806
0.0	47.0102	3024	3.7951	0.5484
0.0	48.0102	3087	2.7604	0.6774
0.0	49.0102	3150	4.3949	0.5484
0.0	50.0102	3213	2.8947	0.6774
0.0	51.0102	3276	4.2413	0.5161
0.1268	52.0102	3339	2.3339	0.7097
0.0	53.0102	3402	3.4769	0.6129
0.0	54.0102	3465	3.5142	0.5806
0.0	55.0102	3528	3.5718	0.5161
0.0036	56.0102	3591	4.1867	0.4839
0.0026	57.0102	3654	2.7411	0.6452
0.0	58.0102	3717	4.0464	0.5484
0.0001	59.0102	3780	3.6255	0.5806
0.0	60.0102	3843	4.7292	0.5161
0.1406	61.0102	3906	3.9876	0.5806
0.0	62.0102	3969	3.4099	0.6129
0.0	63.0102	4032	3.2674	0.5806
0.0	64.0102	4095	3.9749	0.5806
0.0	65.0102	4158	3.3262	0.6129
0.0	66.0102	4221	2.5556	0.7097
0.2639	67.0102	4284	3.6954	0.6129
0.0011	68.0102	4347	3.2776	0.5806
0.0	69.0102	4410	3.6620	0.5806
0.0	70.0102	4473	3.5887	0.5806
0.0	71.0102	4536	4.5040	0.5484
0.0	72.0102	4599	3.8666	0.5484
0.0	73.0102	4662	4.0017	0.5484
0.0	74.0102	4725	3.9422	0.5484
0.0001	75.0102	4788	4.5397	0.5484
0.0	76.0102	4851	3.8405	0.5806
0.0	77.0102	4914	3.9992	0.5806
0.0	78.0102	4977	3.9722	0.5806
0.0	79.0102	5040	3.9421	0.5806
0.2333	80.0102	5103	4.0817	0.5484
0.0	81.0102	5166	3.6669	0.6129
0.0	82.0102	5229	3.6607	0.6129
0.0	83.0102	5292	3.6873	0.6129
0.0	84.0102	5355	4.5979	0.5484
0.0	85.0102	5418	3.8881	0.5806
0.0	86.0102	5481	4.6014	0.5484
0.0	87.0102	5544	3.7988	0.6129
0.0	88.0102	5607	3.8047	0.6129
0.0	89.0102	5670	3.8106	0.6129
0.0001	90.0102	5733	4.1468	0.5806
0.0	91.0102	5796	4.3700	0.5484
0.0	92.0102	5859	4.2392	0.5484
0.0	93.0102	5922	4.2085	0.5484
0.0	94.0102	5985	4.2017	0.5806
0.0	95.0102	6048	4.1972	0.5806
0.0	96.0102	6111	4.1925	0.5806
0.0	97.0102	6174	4.1911	0.5806
0.0	98.0042	6200	4.1910	0.5806

Framework versions

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