learning_rate: 0.0001
train_batch_size: 16
eval_batch_size: 16
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 5814
num_epochs: 30
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Per	Accuracy	Precision	Recall	F1 Macro
2.2446	1.0	728	0.8981	0.2141	0.7830	0.3201	0.3180	0.3180
1.0440	2.0	1456	0.8434	0.1817	0.7966	0.3037	0.2920	0.2969
0.7451	3.0	2184	0.9318	0.1753	0.7994	0.3067	0.2799	0.2916
0.5879	4.0	2912	0.9534	0.1814	0.7960	0.3042	0.2804	0.2911
0.5990	5.0	3640	1.0692	0.1889	0.7897	0.2940	0.2734	0.2830
0.5720	6.0	4368	0.9795	0.1973	0.7816	0.2950	0.2780	0.2862
0.6351	7.0	5096	0.9785	0.2009	0.7811	0.2980	0.2899	0.2936
0.6042	8.0	5824	1.1880	0.2160	0.7700	0.2877	0.2769	0.2821

omarZACK/mdd-arabic-e3-xlsr1b-unannotated-beam

mdd-arabic-e3-xlsr1b-unannotated-beam