beit-base-patch16-224_alpha0.5_temp5.0_t3

This model is a fine-tuned version of c14kevincardenas/ClimBEiT-t3 on the c14kevincardenas/beta_caller_284_person_crop_seq_withlimb dataset. It achieves the following results on the evaluation set:

• Loss: 0.5918
• Accuracy: 0.8482

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: 32
• eval_batch_size: 32
• 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
0.8134	1.0	164	1.4051	0.2549
0.7231	2.0	328	1.1580	0.4978
0.538	3.0	492	0.7405	0.7581
0.4444	4.0	656	0.7384	0.7766
0.3712	5.0	820	0.6911	0.7809
0.3419	6.0	984	0.6714	0.7939
0.3274	7.0	1148	0.6137	0.8167
0.3137	8.0	1312	0.6484	0.8091
0.3083	9.0	1476	0.6518	0.8156
0.2918	10.0	1640	0.6408	0.8102
0.2918	11.0	1804	0.6470	0.8308
0.2915	12.0	1968	0.6152	0.8297
0.2814	13.0	2132	0.6169	0.8232
0.2879	14.0	2296	0.5981	0.8341
0.2865	15.0	2460	0.6006	0.8460
0.2818	16.0	2624	0.5918	0.8482
0.2871	17.0	2788	0.5989	0.8449
0.2757	18.0	2952	0.5924	0.8503
0.282	19.0	3116	0.5949	0.8460
0.2753	20.0	3280	0.5944	0.8492

Framework versions

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