beit-base-finetuned-ade-640-640_alpha0.5_temp3.0_t2

This model is a fine-tuned version of c14kevincardenas/ClimBEiT-t2 on the c14kevincardenas/beta_caller_284_person_crop_seq_withlimb dataset. It achieves the following results on the evaluation set:

• Loss: 0.4847
• Accuracy: 0.8379

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
1.2201	1.0	180	1.3453	0.3804
0.7455	2.0	360	0.8167	0.6897
0.5498	3.0	540	0.6835	0.7609
0.4331	4.0	720	0.6307	0.7737
0.3415	5.0	900	0.5956	0.7915
0.2533	6.0	1080	0.5516	0.8113
0.2366	7.0	1260	0.5602	0.8073
0.2326	8.0	1440	0.6457	0.7866
0.2331	9.0	1620	0.5528	0.8024
0.2284	10.0	1800	0.5510	0.8083
0.1929	11.0	1980	0.5410	0.8152
0.1819	12.0	2160	0.5152	0.8251
0.1677	13.0	2340	0.5095	0.8192
0.1647	14.0	2520	0.4979	0.8310
0.1669	15.0	2700	0.5143	0.8370
0.1511	16.0	2880	0.4890	0.8449
0.1523	17.0	3060	0.4996	0.8360
0.1562	18.0	3240	0.4913	0.8350
0.1474	19.0	3420	0.4876	0.8379
0.1422	20.0	3600	0.4847	0.8379

Framework versions

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