videomae-large

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: 1.5042
• Accuracy: 0.4286

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: 16
• eval_batch_size: 16
• 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: 220

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.6619	0.03	7	2.7017	0.0
2.6232	1.03	14	2.6628	0.0
2.381	2.03	21	2.5798	0.1667
2.2215	3.03	28	2.4757	0.1667
1.7389	4.03	35	2.3636	0.2333
1.3366	5.03	42	2.2424	0.3
1.1946	6.03	49	2.1675	0.3
0.6809	7.03	56	2.0548	0.3667
0.5255	8.03	63	2.0410	0.4
0.3285	9.03	70	1.9539	0.4
0.2849	10.03	77	1.8536	0.4667
0.1832	11.03	84	1.8293	0.4333
0.1307	12.03	91	1.8200	0.4
0.0901	13.03	98	1.8355	0.4
0.0636	14.03	105	1.8201	0.4333
0.0413	15.03	112	1.7750	0.4667
0.0427	16.03	119	1.7460	0.5333
0.0254	17.03	126	1.7804	0.5333
0.0203	18.03	133	1.8869	0.4333
0.0174	19.03	140	1.7741	0.5667
0.0154	20.03	147	1.7401	0.5333
0.0136	21.03	154	1.7672	0.5
0.0116	22.03	161	1.7793	0.5333
0.0123	23.03	168	1.8018	0.4667
0.0102	24.03	175	1.8024	0.5
0.0103	25.03	182	1.8058	0.5
0.0089	26.03	189	1.8106	0.5
0.0088	27.03	196	1.8029	0.5
0.0092	28.03	203	1.7961	0.5
0.0083	29.03	210	1.7940	0.5
0.0099	30.03	217	1.7922	0.5
0.0085	31.01	220	1.7920	0.5

Framework versions

• Transformers 4.37.2
• Pytorch 2.1.0+cu121
• Datasets 2.17.1
• Tokenizers 0.15.2