Unable to change the max_length of captions to train for another dataset #74
Comments
|
I think "max_length" is the only variable to change in prepro.py..... When u get your features successfully... then u run train.py.... now in train.py change parameter "n_time_steps" = new_max_len(in your case 100)+1... and in solver.py you also need to change 3 to 4 parameters. because previously its been hardcoded as per 15 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hey!
I am trying to train the model on another dataset which has relatively longer captions. When I changed max_length = 100 (earlier 15) in main function of prepro.py, it gives the following error:
File "train.py", line 25, in
main()
File "train.py", line 22, in main
solver.train()
File "/home/ubuntu/psk/show-attend-tell/show-attend-and-tell-tensorflow/core/solver.py", line 138, in train
_, l = sess.run([train_op, loss], feed_dict)
File "/home/ubuntu/anaconda3/envs/python2/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 900, in run
run_metadata_ptr)
File "/home/ubuntu/anaconda3/envs/python2/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 1111, in _run
str(subfeed_t.get_shape())))
ValueError: Cannot feed value of shape (128, 102) for Tensor u'Placeholder_1:0', which has shape '(?, 17)'
Can you please help me navigate through it?
TIA! :)
The text was updated successfully, but these errors were encountered: