skywalker290/videomae-vivit-d1

Timesformers-d1

This model is a fine-tuned version of google/vivit-b-16x2-kinetics400 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.7694
• Accuracy: 0.7438

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: 1
• eval_batch_size: 1
• 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: 12010
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.0539	0.1	1201	2.3236	0.6307
0.5697	1.1	2402	1.9547	0.6739
0.5417	2.1	3603	1.7376	0.6951
0.0014	3.1	4804	1.8078	0.6920
1.1162	4.1	6005	1.7942	0.6921
0.0009	5.1	7206	1.4165	0.7779
0.0053	6.1	8407	1.7419	0.7540
1.4804	7.1	9608	1.5797	0.7424
0.6189	8.1	10809	1.9191	0.7305
0.0009	9.1	12010	1.7694	0.7438

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.3