2nzi/videomae-surf-analytics-v3

videomae-surf-analytics-runpod4

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: 0.7259
• Accuracy: 0.9016
• F1: 0.9021

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: 8
• eval_batch_size: 8
• 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: 2760

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
1.2463	0.0337	93	1.2389	0.4344	0.2949
1.024	1.0337	186	1.0343	0.5492	0.5254
0.6932	2.0337	279	0.8280	0.6639	0.6360
0.4467	3.0337	372	0.7665	0.7459	0.7338
0.4449	4.0337	465	0.8715	0.7131	0.6741
0.1371	5.0337	558	1.0560	0.7295	0.7158
0.1789	6.0337	651	0.8218	0.7869	0.7877
0.2125	7.0337	744	0.7612	0.7869	0.7812
0.1561	8.0337	837	0.6051	0.8525	0.8498
0.2297	9.0337	930	0.6321	0.8770	0.8766
0.0692	10.0337	1023	0.7128	0.8443	0.8455
0.0495	11.0337	1116	0.7738	0.8361	0.8353
0.1059	12.0337	1209	0.6213	0.8525	0.8524
0.1672	13.0337	1302	0.7888	0.8443	0.8409
0.0178	14.0337	1395	0.6488	0.8689	0.8658
0.0165	15.0337	1488	0.6845	0.8770	0.8773
0.0166	16.0337	1581	0.8649	0.8525	0.8445
0.0014	17.0337	1674	0.7866	0.8525	0.8516
0.0473	18.0337	1767	0.6390	0.8770	0.8776
0.0441	19.0337	1860	0.8235	0.8361	0.8342
0.0006	20.0337	1953	0.6014	0.8852	0.8856
0.0005	21.0337	2046	0.7581	0.8689	0.8672
0.0032	22.0337	2139	0.6454	0.8770	0.8772
0.0565	23.0337	2232	0.8096	0.8525	0.8542
0.011	24.0337	2325	0.6807	0.8852	0.8858
0.0146	25.0337	2418	0.7754	0.8689	0.8696
0.0004	26.0337	2511	0.7246	0.8852	0.8857
0.0004	27.0337	2604	0.7165	0.8934	0.8942
0.0003	28.0337	2697	0.7232	0.9016	0.9021
0.0177	29.0228	2760	0.7259	0.9016	0.9021

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.1+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1