CHAR_VECTOR is correct?
yoyoyo-yo opened this issue · comments
Thank you for your great job!
I tried your pretrained model, but got error.
Traceback (most recent call last):
File "/Users/yoshitonagaoka/dev/miniconda3/envs/deepscholar-crnn/lib/python3.7/site-packages/tensorflow/python/client/session.py", line 1356, in _do_call
return fn(*args)
File "/Users/yoshitonagaoka/dev/miniconda3/envs/deepscholar-crnn/lib/python3.7/site-packages/tensorflow/python/client/session.py", line 1341, in _run_fn
options, feed_dict, fetch_list, target_list, run_metadata)
File "/Users/yoshitonagaoka/dev/miniconda3/envs/deepscholar-crnn/lib/python3.7/site-packages/tensorflow/python/client/session.py", line 1429, in _call_tf_sessionrun
run_metadata)
tensorflow.python.framework.errors_impl.InvalidArgumentError: Assign requires shapes of both tensors to match. lhs shape= [512,71] rhs shape= [512,70]
[[{{node save/Assign_1}}]]
I fixed CHAR_VECTOR in run.py like below, and worked well.
#CHAR_VECTOR = "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ-'.!?,"&"
CHAR_VECTOR = "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ-'.!?,&"
Please tell the correct CHAR_VECTOR for your pretrained model.
Please see the issue #61