VideoMAE_WLASL_100_200_epochs_p20_SR_8_again

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

• Loss: 3.6313
• Top 1 Accuracy: 0.4408
• Top 5 Accuracy: 0.7456
• Top 10 Accuracy: 0.8254
• Accuracy: 0.4408
• Precision: 0.5153
• Recall: 0.4408
• F1: 0.4320

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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: 36000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Top 1 Accuracy	Top 5 Accuracy	Top 10 Accuracy	Accuracy	Precision	Recall	F1
18.7179	0.005	180	4.6606	0.0118	0.0503	0.1006	0.0118	0.0008	0.0118	0.0015
18.5895	1.0050	360	4.6279	0.0266	0.0740	0.1065	0.0266	0.0048	0.0266	0.0055
18.4949	2.0050	540	4.6162	0.0296	0.0858	0.1450	0.0296	0.0040	0.0296	0.0068
18.4563	3.0050	721	4.6074	0.0325	0.0740	0.1124	0.0325	0.0026	0.0325	0.0046
18.4839	4.005	901	4.6099	0.0237	0.0799	0.1243	0.0237	0.0012	0.0237	0.0022
18.2388	5.0050	1081	4.6018	0.0237	0.0710	0.1450	0.0237	0.0017	0.0237	0.0030
18.2327	6.0050	1261	4.6311	0.0207	0.0621	0.1213	0.0207	0.0035	0.0207	0.0057
18.146	7.0050	1442	4.6073	0.0148	0.0917	0.1568	0.0148	0.0014	0.0148	0.0026
17.979	8.005	1622	4.6589	0.0148	0.0858	0.1302	0.0148	0.0009	0.0148	0.0017
17.7095	9.0050	1802	4.5615	0.0237	0.1006	0.1716	0.0237	0.0052	0.0237	0.0070
17.1282	10.0050	1982	4.3098	0.0473	0.1509	0.2485	0.0473	0.0101	0.0473	0.0115
16.1191	11.0050	2163	4.0549	0.0444	0.2544	0.3846	0.0444	0.0253	0.0444	0.0200
14.7815	12.005	2343	3.7599	0.1095	0.3402	0.5296	0.1095	0.0711	0.1095	0.0726
13.6312	13.0050	2523	3.5324	0.1509	0.4320	0.5947	0.1509	0.0861	0.1509	0.0937
11.9507	14.0050	2703	3.2292	0.2101	0.5444	0.7130	0.2101	0.1555	0.2101	0.1575
10.5467	15.0050	2884	3.0303	0.2367	0.5562	0.7219	0.2367	0.1863	0.2367	0.1805
8.8902	16.005	3064	2.7703	0.3491	0.6479	0.8077	0.3491	0.3817	0.3491	0.3127
7.4374	17.0050	3244	2.5973	0.3905	0.7130	0.8136	0.3905	0.3510	0.3905	0.3365
6.1127	18.0050	3424	2.3462	0.4320	0.7337	0.8580	0.4320	0.4674	0.4320	0.4055
5.1867	19.0050	3605	2.3088	0.4201	0.7515	0.8639	0.4201	0.4386	0.4201	0.4045
4.0661	20.005	3785	2.3607	0.3876	0.7189	0.8402	0.3905	0.4236	0.3905	0.3654
3.1864	21.0050	3965	2.2799	0.4201	0.7574	0.8462	0.4201	0.4789	0.4201	0.3998
2.5534	22.0050	4145	2.2289	0.4260	0.7249	0.8669	0.4260	0.4547	0.4260	0.4115
1.8595	23.0050	4326	2.0805	0.4467	0.7899	0.8728	0.4467	0.4872	0.4467	0.4271
1.6117	24.005	4506	2.0355	0.4793	0.7781	0.8846	0.4793	0.5106	0.4793	0.4609
1.1388	25.0050	4686	2.0650	0.5178	0.7663	0.8639	0.5178	0.5759	0.5178	0.5051
0.8489	26.0050	4866	1.9968	0.5178	0.7811	0.8787	0.5178	0.5630	0.5178	0.5011
0.7129	27.0050	5047	2.1062	0.5296	0.7663	0.8639	0.5296	0.5699	0.5296	0.5064
0.6167	28.005	5227	2.2424	0.4822	0.7781	0.8669	0.4822	0.5333	0.4822	0.4601
0.5676	29.0050	5407	2.3891	0.4408	0.7485	0.8580	0.4408	0.5477	0.4408	0.4309
0.4871	30.0050	5587	2.1354	0.4882	0.8136	0.9172	0.4911	0.5456	0.4911	0.4713
0.3447	31.0050	5768	2.1996	0.5118	0.8107	0.8787	0.5118	0.5613	0.5118	0.4931
0.358	32.005	5948	2.2036	0.5178	0.8136	0.8964	0.5178	0.5841	0.5178	0.5025
0.1964	33.0050	6128	2.2685	0.5059	0.8047	0.9053	0.5059	0.5579	0.5059	0.4902
0.2698	34.0050	6308	2.3974	0.5089	0.7959	0.8669	0.5089	0.5442	0.5089	0.4919
0.2084	35.0050	6489	2.5295	0.5148	0.8136	0.8550	0.5148	0.5617	0.5148	0.4948
0.2268	36.005	6669	2.3980	0.5385	0.8195	0.8728	0.5385	0.5945	0.5385	0.5275
0.1354	37.0050	6849	2.4901	0.5592	0.7959	0.8669	0.5592	0.6243	0.5592	0.5483
0.1738	38.0050	7029	2.6473	0.5178	0.7722	0.8757	0.5178	0.5754	0.5178	0.5004
0.1795	39.0050	7210	2.5686	0.5414	0.7929	0.8876	0.5414	0.6179	0.5414	0.5320
0.1308	40.005	7390	2.5885	0.5089	0.8018	0.8787	0.5089	0.5755	0.5089	0.5035
0.1935	41.0050	7570	2.8128	0.4970	0.7988	0.8787	0.4970	0.5563	0.4970	0.4846
0.1201	42.0050	7750	2.6633	0.5355	0.8018	0.8757	0.5355	0.5693	0.5355	0.5190
0.1912	43.0050	7931	2.8220	0.5059	0.7811	0.8580	0.5059	0.5849	0.5059	0.4903
0.18	44.005	8111	3.0467	0.4734	0.7456	0.8462	0.4734	0.5019	0.4734	0.4495
0.2313	45.0050	8291	2.6114	0.5503	0.8077	0.8669	0.5503	0.5798	0.5503	0.5284
0.0689	46.0050	8471	2.8944	0.5148	0.7811	0.8580	0.5148	0.5561	0.5148	0.4981
0.0908	47.0050	8652	2.8956	0.5355	0.7692	0.8669	0.5355	0.5728	0.5355	0.5168
0.0942	48.005	8832	2.8800	0.5148	0.7781	0.8402	0.5148	0.5448	0.5148	0.4965
0.2255	49.0050	9012	3.1947	0.4941	0.7633	0.8343	0.4941	0.5324	0.4941	0.4762
0.1361	50.0050	9192	2.8410	0.5385	0.7663	0.8609	0.5385	0.5815	0.5385	0.5195
0.1301	51.0050	9373	3.1655	0.5089	0.7604	0.8491	0.5118	0.5529	0.5118	0.4932
0.1755	52.005	9553	3.6231	0.4349	0.7219	0.8166	0.4349	0.4579	0.4349	0.4119
0.1833	53.0050	9733	3.1796	0.5059	0.7781	0.8521	0.5059	0.5421	0.5059	0.4865
0.2115	54.0050	9913	3.2417	0.4970	0.7574	0.8343	0.4970	0.5495	0.4970	0.4810
0.2306	55.0050	10094	3.1876	0.4911	0.7367	0.8343	0.4911	0.5179	0.4911	0.4726
0.119	56.005	10274	3.2365	0.4941	0.7367	0.8284	0.4941	0.5319	0.4941	0.4723
0.1542	57.0050	10454	3.6313	0.4408	0.7456	0.8254	0.4408	0.5153	0.4408	0.4320

Framework versions

• Transformers 4.46.1
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.1