c14kevincardenas/beit-base-patch16-384_alpha0.7_temp3.0_t2

beit-base-patch16-384_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: 0.5082
• Accuracy: 0.8409

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
1.1161	1.0	90	1.3958	0.3034
0.8866	2.0	180	0.9934	0.5830
0.5473	3.0	270	0.6915	0.7451
0.4068	4.0	360	0.6147	0.7836
0.2803	5.0	450	0.5613	0.8132
0.232	6.0	540	0.5583	0.8073
0.2132	7.0	630	0.5762	0.8103
0.1911	8.0	720	0.5657	0.8192
0.1705	9.0	810	0.5529	0.8103
0.1603	10.0	900	0.5434	0.8172
0.1553	11.0	990	0.5209	0.8221
0.1475	12.0	1080	0.5409	0.8182
0.1446	13.0	1170	0.5163	0.8340
0.1316	14.0	1260	0.5329	0.8251
0.1317	15.0	1350	0.5183	0.8389
0.1284	16.0	1440	0.5082	0.8409
0.1221	17.0	1530	0.5179	0.8389
0.1204	18.0	1620	0.5164	0.8370
0.1192	19.0	1710	0.5117	0.8439
0.1167	20.0	1800	0.5123	0.8429

Framework versions

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