I am new to machine learning and trying to use the tool to train for creating 2D game maps. The format of my maps are 20 rows by 100 columns of characters with the rows and columns rotated 90 degrees. where each column becomes a 20 character word with words separated by spaces. For training I did
but the biggest problem I am having is that the results are not always 20 character words (even though the training data has 20 character words for every word) and as a result my sampled level is not readable.
I am new to machine learning and trying to use the tool to train for creating 2D game maps. The format of my maps are 20 rows by 100 columns of characters with the rows and columns rotated 90 degrees. where each column becomes a 20 character word with words separated by spaces. For training I did
python train.py --data_dir data\levels --seq_length 10 --batch_size 20 --num_epochs 1000
and for sampling I did
python sample.py --prime 4```````````````````
but the biggest problem I am having is that the results are not always 20 character words (even though the training data has 20 character words for every word) and as a result my sampled level is not readable.