fall-detection-ft

This model is a fine-tuned version of popkek00/trainer_output on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5574
• Accuracy: 0.7395
• F1: 0.8437

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: 2e-05
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• optimizer: Use adamw_torch_fused 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
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.5734	1.0	42	0.5955	0.7111	0.8310
0.5998	2.0	84	0.6113	0.6939	0.8116
0.5725	3.0	126	0.5574	0.7395	0.8437
0.5145	4.0	168	0.5487	0.7463	0.8371
0.5427	5.0	210	0.5389	0.7380	0.8344
0.499	6.0	252	0.5532	0.7246	0.8045
0.4837	7.0	294	0.6000	0.6826	0.7599
0.4965	8.0	336	0.5712	0.7081	0.7921
0.4505	9.0	378	0.6617	0.6347	0.7182
0.4533	10.0	420	0.7154	0.6168	0.6784
0.4588	11.0	462	0.6455	0.6549	0.7331
0.4251	12.0	504	0.6885	0.6093	0.6713
0.4143	13.0	546	0.6569	0.6265	0.7007
0.4031	14.0	588	0.7144	0.5936	0.6574
0.4011	15.0	630	0.6977	0.6033	0.6696

Framework versions

• Transformers 4.57.6
• Pytorch 2.10.0+cu128
• Datasets 4.8.4
• Tokenizers 0.22.2