Florence-2-FT-JP-OCR

This model is a fine-tuned version of microsoft/Florence-2-base-ft on EtashGuha/JapaneseDocQA dataset. It achieves the following results on the evaluation set:

• Loss: 2.5827

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: 5e-06
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss
14.3012	0.9662	100	3.1360
12.8837	1.9275	200	2.9369
12.1693	2.8889	300	2.7978
11.6903	3.8502	400	2.7166
11.3443	4.8116	500	2.6688
11.1585	5.7729	600	2.6340
10.9682	6.7343	700	2.6140
10.9556	7.6957	800	2.5960
10.7355	8.6570	900	2.5882
10.6794	9.6184	1000	2.5827

Framework versions

• Transformers 4.48.0.dev0
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0