rnnsynth stalls out #5
Comments
|
Do you have verbose output turned on either in .save file verbose=true or in command line --verbose=true? |
|
@szcom Hmm, I just checked and looks like verbose is set to false. Would setting verbose to true hang things up? I'm starting to think there is nothing in the sequences that it's trying to load, so it gets there and looks for things but never finds any and just sits waiting... I'm guessing this fails somewhere in the load_sequences function in the NetcdfDataset.hpp between line 338 and 341 because I'm never seeing any console output after |
|
@aferriss i was suggesting to use verbose=true to see if there is anything else being printed out or it is the end of it. You can play with --dataFraction=.001 to skip loading all of the dataset is it possible to attach gdb and see the stack trace? may be some clues there |
|
@szcom Ah well I feel a little foolish now. I didn't realize I had to pass the test string in via stdin when running rnnsynth. Still not working properly though. If I do --sentence="some sentence" it just hangs like before. But If I do It loads my model and seems to start working but then... And then it quits. I'm not that familiar with gdb, but I'll give that a try to see if it sheds light on anything. Would it be like: ctrl-c at the hang and then edit: I just did what I wrote down about for gdb and it spat this out for me |
|
under gdb it dies if you stop on a breakpoint and try to continue... |
|
what you need is to write your sentence on the screen after line |
|
@melserafy ah that was it, I feel silly now. Didn't realize it was waiting for input. Thanks! |
|
@aferriss did the rnnsynth give the right sentence output |
|
@melserafy I'm not entirely sure because I didn't train the network for the full amount of epochs so I just get some weird jibberish. It didn't seem like it was giving the right amount of samples however. For instance: Why would "test test" generate fewer samples than just "test"? If I run "test test" again I get a longer one, but it seems to alternate between shorter and longer sample outputs. |
|
when running rnnsynth how do you save the output? |
|
@andkylrob you can just send it to a text file. rnnsynth is just printing to stdout anyway. so:
should work, but keep in mind you won't see any output on the terminal since it's being sent to a file. You could see both by piping to tee I think
|
when I run rnnsynth I get this output that looks like it's loading all the weights from my model, but then it starts to load sequences and just hangs. I looked into activity monitor and rnnsynth drops down to 0% cpu. Here's my output:
It never moves past this point. I didn't let my networks train until 20 epochs, but would that make a difference here? Does rnnsynth need any additional input parameters? I did notice this line during the CMAKE build process but assumed it was fine with the library that it found
The text was updated successfully, but these errors were encountered: