khmer-trocr-base-printed

This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4627
• Cer: 0.7359

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 4e-05
• train_batch_size: 8
• eval_batch_size: 6
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
1.0771	0.2612	1000	1.0644	0.8631
1.0182	0.5223	2000	0.9849	0.8791
0.9409	0.7835	3000	0.9387	0.8704
0.9053	1.0447	4000	0.8917	0.8646
0.8428	1.3058	5000	0.8280	0.8600
0.729	1.5670	6000	0.7200	0.8192
0.6011	1.8282	7000	0.6203	0.7893
0.5268	2.0893	8000	0.5568	0.7621
0.4819	2.3505	9000	0.5150	0.7554
0.4362	2.6116	10000	0.4799	0.7412
0.4555	2.8728	11000	0.4627	0.7359

Framework versions

• Transformers 4.42.3
• Pytorch 2.1.2
• Datasets 2.20.0
• Tokenizers 0.19.1