beit-base-patch16-384_alpha0.7_temp3.0_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: 0.7145
• Accuracy: 0.8210

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
0.5651	1.0	164	1.4176	0.2581
0.513	2.0	328	1.2863	0.4642
0.3848	3.0	492	0.9326	0.6779
0.3243	4.0	656	0.8487	0.7462
0.2955	5.0	820	0.7188	0.8124
0.2665	6.0	984	0.7605	0.8059
0.2592	7.0	1148	0.7145	0.8210
0.256	8.0	1312	0.7547	0.7983
0.2509	9.0	1476	0.7419	0.8080
0.2361	10.0	1640	0.7513	0.8113
0.2333	11.0	1804	0.7794	0.7972
0.2365	12.0	1968	0.7325	0.8178
0.2271	13.0	2132	0.7417	0.8189
0.2317	14.0	2296	0.7332	0.8156
0.2316	15.0	2460	0.7399	0.8102
0.2265	16.0	2624	0.7357	0.8134
0.2304	17.0	2788	0.7330	0.8189
0.2201	18.0	2952	0.7288	0.8145
0.2253	19.0	3116	0.7330	0.8200
0.2197	20.0	3280	0.7306	0.8221

Framework versions

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