c14kevincardenas/mobilevit-small_alpha0.7_temp5.0_t2

mobilevit-small_alpha0.7_temp5.0

This model is a fine-tuned version of c14kevincardenas/ClimBEiTv2 on the c14kevincardenas/beta_caller_284_person_crop_seq_withlimb_2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.8550
• Accuracy: 0.6680

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: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.1311	1.0	90	1.4279	0.2470
1.0789	2.0	180	1.3331	0.3192
0.9867	3.0	270	1.2525	0.3864
0.8433	4.0	360	1.0671	0.4970
0.7496	5.0	450	0.9999	0.5603
0.6948	6.0	540	0.9374	0.5988
0.6495	7.0	630	0.9065	0.6146
0.6101	8.0	720	0.9331	0.6008
0.5591	9.0	810	0.8673	0.6571
0.5406	10.0	900	0.8705	0.6235
0.5188	11.0	990	0.8579	0.6630
0.4875	12.0	1080	0.8809	0.6324
0.4596	13.0	1170	0.8904	0.6324
0.4418	14.0	1260	0.8596	0.6542
0.4385	15.0	1350	0.8769	0.6304
0.4175	16.0	1440	0.8625	0.6561
0.413	17.0	1530	0.8578	0.6571
0.3973	18.0	1620	0.8573	0.6581
0.3964	19.0	1710	0.8550	0.6680
0.4004	20.0	1800	0.8711	0.6403

Framework versions

• Transformers 4.45.2
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1