c14kevincardenas/deit-small-distilled-patch16-224_alpha0.7_temp3.0_t2

deit-small-distilled-patch16-224_alpha0.7_temp3.0

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: 0.7187
• Accuracy: 0.7292

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
1.1503	1.0	90	1.4292	0.2717
1.0007	2.0	180	1.2262	0.4526
0.6862	3.0	270	0.8957	0.6186
0.4297	4.0	360	0.7742	0.6927
0.2549	5.0	450	0.7670	0.7006
0.1769	6.0	540	0.8104	0.6848
0.1607	7.0	630	0.7589	0.7243
0.1481	8.0	720	0.7446	0.7174
0.1418	9.0	810	0.7526	0.7134
0.1308	10.0	900	0.7379	0.7204
0.1246	11.0	990	0.7421	0.7283
0.1226	12.0	1080	0.7369	0.7184
0.1143	13.0	1170	0.7189	0.7332
0.1125	14.0	1260	0.7187	0.7292
0.1109	15.0	1350	0.7294	0.7292
0.1087	16.0	1440	0.7224	0.7213
0.1075	17.0	1530	0.7298	0.7243
0.1015	18.0	1620	0.7258	0.7253
0.1042	19.0	1710	0.7268	0.7273
0.1014	20.0	1800	0.7265	0.7263

Framework versions

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