learning_rate: 5e-05
train_batch_size: 16
eval_batch_size: 8
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: 500
num_epochs: 10
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Cer
5.2955	0.5272	300	1.0741	0.2438
1.8409	1.0545	600	0.9351	0.2243
1.4477	1.5817	900	0.8612	0.1868
1.1644	2.1090	1200	0.8993	0.1946
0.9022	2.6362	1500	0.8686	0.1900
0.7129	3.1634	1800	0.9532	0.1820
0.5254	3.6907	2100	0.9420	0.1823
0.3952	4.2179	2400	1.1475	0.1815
0.2816	4.7452	2700	1.1638	0.1800
0.1984	5.2724	3000	1.3505	0.1817
0.1499	5.7996	3300	1.4472	0.1801
0.1010	6.3269	3600	1.5881	0.1816
0.0793	6.8541	3900	1.6870	0.1768
0.0569	7.3814	4200	1.7939	0.1799
0.0393	7.9086	4500	1.9141	0.1764
0.0251	8.4359	4800	2.0075	0.1761
0.0172	8.9631	5100	2.0690	0.1768
0.0097	9.4903	5400	2.1780	0.1744

dhasmana/Chhattisgarhi-Magadhi-Marathi-Rajasthani-Kumaoni-w2v-bert-2.0

Chhattisgarhi-Magadhi-Marathi-Rajasthani-Kumaoni-w2v-bert-2.0