Update README.md

README.md

@@ -34,6 +34,6 @@ For a demo of the model, head over to the Hugging Face Spaces for the **[Netherf
 ## Training
 
 The pre-trained model [t5-base-dutch](https://huggingface.co/flax-community/t5-base-dutch) was fine-tuned with a constant learning rate of 0.0005 and a batch size of 64 for 10.000 steps. 
-The performance of this model can be improved with longer training. Unfortunately due to a bug, an earlier training script would not save intermediate checkpoints, and had been started for 6 epochs, and would finish past the availability of the TPU-VM. Since there was limited time left, the fine-tuning was restarted without evaluation and for only half an epoch (10.000 steps).
+The performance of this model can be improved with longer training. Unfortunately due to a bug, an earlier training script would not save intermediate checkpoints and had been started for 6 epochs, which would have it finish past the TPU-VM availability schedule. Since there was limited time left, the fine-tuning was restarted without evaluation and for only half an epoch (10.000 steps).