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
5.1478	0.4488	300	1.0622	0.2308
1.6844	0.8975	600	0.9067	0.2129
1.2921	1.3455	900	0.8482	0.2082
1.0521	1.7943	1200	0.8565	0.1979
0.8166	2.2423	1500	0.8877	0.1917
0.6523	2.6911	1800	0.9594	0.1926
0.5035	3.1391	2100	1.0944	0.1931
0.3788	3.5879	2400	1.1248	0.1933
0.2940	4.0359	2700	1.3119	0.1902
0.2035	4.4847	3000	1.3604	0.1928
0.1769	4.9334	3300	1.4865	0.1883
0.1214	5.3815	3600	1.5188	0.1919
0.1023	5.8302	3900	1.7170	0.1874
0.0772	6.2782	4200	1.8157	0.1857
0.0617	6.7270	4500	1.8770	0.1871
0.0473	7.1750	4800	2.0192	0.1864
0.0330	7.6238	5100	2.0650	0.1849
0.0270	8.0718	5400	2.1796	0.1857
0.0172	8.5206	5700	2.3015	0.1839
0.0144	8.9693	6000	2.3617	0.1834
0.0086	9.4174	6300	2.4157	0.1833
0.0052	9.8661	6600	2.4368	0.1826

dhasmana/Kumaoni-Bhojpuri-Chhattisgarhi-Angika-Maithili-Magadhi-Konkani-w2v-bert-2.0

Kumaoni-Bhojpuri-Chhattisgarhi-Angika-Maithili-Magadhi-Konkani-w2v-bert-2.0