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
2.8418	0.5133	300	1.1076	0.2533
0.9814	1.0257	600	0.9325	0.2115
0.7663	1.5389	900	0.8468	0.1985
0.6216	2.0513	1200	0.8789	0.1888
0.4946	2.5646	1500	0.8595	0.1865
0.406	3.0770	1800	0.9114	0.1908
0.3135	3.5902	2100	0.9660	0.1813
0.2505	4.1027	2400	0.9396	0.1868
0.1816	4.6159	2700	0.9880	0.1827
0.1402	5.1283	3000	1.1988	0.1799
0.1026	5.6416	3300	1.2953	0.1765
0.0776	6.1540	3600	1.3251	0.1818
0.0555	6.6672	3900	1.3804	0.1777
0.0421	7.1796	4200	1.3845	0.1797
0.0282	7.6929	4500	1.5102	0.1761
0.0216	8.2053	4800	1.5766	0.1785
0.0128	8.7186	5100	1.7343	0.1758
0.0077	9.2310	5400	1.8764	0.1736
0.0055	9.7442	5700	1.8944	0.1740

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

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