cvt-13-384-in22k-FV-finetuned-memes

This model is a fine-tuned version of microsoft/cvt-13-384-22k on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.5595
• Accuracy: 0.8346
• Precision: 0.8327
• Recall: 0.8346
• F1: 0.8322

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: 0.00012
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• 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	Precision	Recall	F1
1.4066	0.99	20	1.2430	0.5124	0.5141	0.5124	0.4371
1.0813	1.99	40	0.8244	0.6893	0.6834	0.6893	0.6616
0.8392	2.99	60	0.6334	0.7612	0.7670	0.7612	0.7570
0.7065	3.99	80	0.5819	0.7767	0.7799	0.7767	0.7672
0.5751	4.99	100	0.5365	0.8176	0.8216	0.8176	0.8130
0.4896	5.99	120	0.4943	0.8308	0.8257	0.8308	0.8265
0.4487	6.99	140	0.5399	0.8107	0.8069	0.8107	0.8054
0.4349	7.99	160	0.4892	0.8300	0.8285	0.8300	0.8273
0.43	8.99	180	0.4984	0.8454	0.8465	0.8454	0.8426
0.4372	9.99	200	0.5573	0.8192	0.8221	0.8192	0.8157
0.3994	10.99	220	0.5158	0.8300	0.8284	0.8300	0.8281
0.3883	11.99	240	0.5495	0.8354	0.8317	0.8354	0.8314
0.406	12.99	260	0.5298	0.8284	0.8285	0.8284	0.8246
0.3355	13.99	280	0.5401	0.8393	0.8346	0.8393	0.8357
0.395	14.99	300	0.5915	0.8308	0.8278	0.8308	0.8261
0.3612	15.99	320	0.5852	0.8408	0.8378	0.8408	0.8368
0.3765	16.99	340	0.5509	0.8385	0.8351	0.8385	0.8356
0.3688	17.99	360	0.5668	0.8416	0.8398	0.8416	0.8387
0.3503	18.99	380	0.5626	0.8393	0.8371	0.8393	0.8365
0.3611	19.99	400	0.5595	0.8346	0.8327	0.8346	0.8322

Framework versions

• Transformers 4.24.0.dev0
• Pytorch 1.11.0+cu102
• Datasets 2.6.1.dev0
• Tokenizers 0.13.1