timesformer_wlasl_100_20ep_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: 1.4422
• Accuracy: 0.6657

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: 3600
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
18.516	0.05	180	4.3268	0.0680
14.6282	1.0499	360	3.2382	0.3077
9.1983	2.0499	540	2.5109	0.4379
5.217	3.0501	721	2.0461	0.5148
2.6858	4.05	901	1.7717	0.5769
1.2884	5.0499	1081	1.6758	0.6006
0.6468	6.0499	1261	1.5235	0.6213
0.3398	7.0501	1442	1.4498	0.6272
0.1804	8.05	1622	1.4702	0.6272
0.1119	9.0499	1802	1.4163	0.6302
0.079	10.0499	1982	1.4269	0.6538
0.0686	11.0501	2163	1.4097	0.6568
0.0626	12.05	2343	1.3878	0.6657
0.0485	13.0499	2523	1.4528	0.6686
0.0613	14.0499	2703	1.4348	0.6627
0.053	15.0501	2884	1.4080	0.6716
0.0435	16.05	3064	1.4331	0.6627
0.029	17.0499	3244	1.4446	0.6568
0.0292	18.0499	3424	1.4439	0.6686
0.0199	19.0487	3600	1.4422	0.6657

Framework versions

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