timesformer_wlasl_100__signer_30ep_coR

This model is a fine-tuned version of facebook/timesformer-base-finetuned-k400 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.4623
• Accuracy: 0.6479

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: 5400
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
18.6381	0.0333	180	4.4513	0.0385
15.7836	1.0333	360	3.5410	0.2308
10.8669	2.0332	540	2.8053	0.3757
6.7643	3.0334	721	2.2128	0.4704
3.606	4.0333	901	1.8708	0.5562
1.7698	5.0333	1081	1.7562	0.5651
0.8727	6.0332	1261	1.5556	0.6154
0.4877	7.0334	1442	1.5320	0.6154
0.2813	8.0333	1622	1.5887	0.5858
0.1615	9.0333	1802	1.5194	0.6331
0.1182	10.0332	1982	1.4934	0.6420
0.0778	11.0334	2163	1.4369	0.6391
0.0849	12.0333	2343	1.4821	0.6746
0.0747	13.0333	2523	1.5323	0.6479
0.0734	14.0332	2703	1.4419	0.6746
0.0747	15.0334	2884	1.4604	0.6746
0.0506	16.0333	3064	1.4617	0.6716
0.0509	17.0333	3244	1.4623	0.6479

Framework versions

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