git-large-r-coco-IDB2-V2-IDB2-VAtlasv2

This model is a fine-tuned version of ooliverz/git-large-r-coco-IDB2-V2 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.2986

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
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• optimizer: Use adafactor and the args are: No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 500
• num_epochs: 150
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
5.736	5.0	5	2.0136
5.4953	10.0	10	1.9221
5.2338	15.0	15	1.7424
4.2459	20.0	20	1.4205
3.5741	25.0	25	1.2983
3.3273	30.0	30	1.2457
3.1782	35.0	35	1.0753
3.0982	40.0	40	1.0267
2.626	45.0	45	0.9531
2.4711	50.0	50	0.8802
2.4721	55.0	55	0.8395
2.2335	60.0	60	0.7950
2.1829	65.0	65	0.7458
1.976	70.0	70	0.6922
1.957	75.0	75	0.6412
1.7707	80.0	80	0.6177
1.5535	85.0	85	0.5689
1.6445	90.0	90	0.5447
1.5019	95.0	95	0.5097
1.4978	100.0	100	0.4803
1.2429	105.0	105	0.4540
1.1654	110.0	110	0.4323
1.1942	115.0	115	0.4091
1.0655	120.0	120	0.3813
0.9715	125.0	125	0.3681
0.9303	130.0	130	0.3517
0.8991	135.0	135	0.3453
0.8617	140.0	140	0.3263
0.8206	145.0	145	0.3078
0.7037	150.0	150	0.2986

Framework versions

• Transformers 4.46.1
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.20.2