learning_rate: 0.0001
train_batch_size: 32
eval_batch_size: 32
seed: 42
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
num_epochs: 30

Training Loss	Epoch	Step	Accuracy	F1 Macro	Validation Loss	Per	Precision	Recall
0.6854	1.0	728	0.7880	0.2942	0.8199	0.1959	0.2979	0.2935
0.4210	2.0	1456	0.7947	0.2913	0.9654	0.1878	0.3011	0.2828
0.3884	3.0	2184	0.7797	0.2895	0.9815	0.2032	0.2963	0.2831
0.3001	4.0	2912	1.0215	0.2031	0.7756	0.2867	0.2801	0.2828
0.2835	5.0	3640	1.0230	0.2115	0.7759	0.2949	0.2793	0.2869
0.2106	6.0	4368	1.0411	0.2112	0.7705	0.2877	0.2767	0.2819
0.2966	7.0	5096	1.0804	0.2244	0.7718	0.2866	0.2731	0.2796
0.1863	8.0	5824	1.0495	0.2157	0.7657	0.2790	0.2688	0.2736

omarZACK/mdd-arabic-e6-xlsr1b-speed-perturb-beam

mdd-arabic-e6-xlsr1b-speed-perturb-beam