_base_nougat_AHR

This model is a fine-tuned version of facebook/nougat-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.2011

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 6
• total_train_batch_size: 48
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss
2.1218	0.9978	76	2.0613
1.9342	1.9956	152	1.8993
1.835	2.9934	228	1.8198
1.7619	3.9912	304	1.7794
1.7179	4.9891	380	1.7415
1.6621	6.0	457	1.7167
1.6321	6.9978	533	1.6912
1.5972	7.9956	609	1.6770
1.5809	8.9934	685	1.6699
1.5203	9.9912	761	1.6632
1.5007	10.9891	837	1.6380
1.4308	12.0	914	1.6039
1.3866	12.9978	990	1.5650
1.3315	13.9956	1066	1.5164
1.2601	14.9934	1142	1.4441
1.1786	15.9912	1218	1.3988
1.1358	16.9891	1294	1.3069
1.0661	18.0	1371	1.3010
1.0571	18.9978	1447	1.2936
1.0299	19.9956	1523	1.2539
1.0327	20.9934	1599	1.2193
0.9878	21.9912	1675	1.1983
0.9844	22.9891	1751	1.2063
0.9645	24.0	1828	1.2009
0.9645	24.9978	1904	1.2011

Framework versions

• Transformers 4.46.3
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.20.3