learning_rate: 5e-05
train_batch_size: 16
eval_batch_size: 16
seed: 1
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: constant_with_warmup
lr_scheduler_warmup_steps: 2000
training_steps: 20000

Training Loss	Epoch	Step	Validation Loss	Chrf
10.3266	0.3031	2000	3.6628	37.9925
4.1088	0.6062	4000	2.5458	46.6768
2.6206	0.9092	6000	1.8347	56.0621
2.1827	1.2123	8000	1.4675	62.6109
1.8646	1.5154	10000	1.2587	66.9559
1.5054	1.8185	12000	1.1203	69.8883
1.3636	2.1215	14000	1.0030	72.0171
1.2315	2.4246	16000	0.9427	73.3205
1.1395	2.7277	18000	0.8902	74.4683
1.0638	3.0308	20000	0.8543	75.7075

Curiousfox/mT5_tai-lo_to_chinese_ver6.0

outputs