medasr-ghana-v2

This model is a fine-tuned version of google/medasr on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.1764
• Wer: 0.4116

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: 0.0001
• train_batch_size: 4
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 32
• total_train_batch_size: 128
• 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
• lr_scheduler_warmup_steps: 19
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
85.5056	7.1527	100	2.2687	0.5682
71.0330	14.3055	200	1.9314	0.5153
69.0262	21.4582	300	1.8050	0.5015
66.9954	28.6110	400	1.7908	0.4932

Framework versions

• Transformers 5.0.0.dev0
• Pytorch 2.9.1+cu128
• Datasets 2.20.0
• Tokenizers 0.22.2