search

dattalab / pyhsmm-library-models

library models built on top of pyhsmm
0 stars 1 forks source link

Parallel sampling stalls #10

Closed alexbw closed 11 years ago

alexbw commented 11 years ago

No stack trace, because it just hangs. Don't know what to dive in to check this. Training with 1000 samples of 100D data using the master branches of both pyhsmm and pyhsmm-library-models

mattjj commented 11 years ago

Any more info? What hashes are you on? Because it's working fine for me with f825c1f in library-models and dad58a5 on pyhsmm; I just ran it.

mattjj commented 11 years ago

Are you on the new ipython? I'm running 0.13.2

And what's the trace when you ctrl-c it? It will be informative to see where it's hanging.

alexbw commented 11 years ago

I'm on the new IPython, v1.0. Ohhhhhh. That definitely could be an issue here. Dagnubbit, I was really liking the new IPython notebook... I'll run an example script soon.

On Sun, Jul 28, 2013 at 8:07 PM, Matthew Johnson notifications@github.comwrote:

Are you on the new ipython? I'm running 0.13.2

— Reply to this email directly or view it on GitHubhttps://github.com/dattalab/pyhsmm-library-models/issues/10#issuecomment-21694119 .

mattjj commented 11 years ago

I can upgrade and figure it out; if that's the issue I'm sure it'll be easy to fix. But check the hashes of the commits you're on, too.

mattjj commented 11 years ago

Everything is working the same on my user on Jefferson. If you get a chance, please try these things:

  1. rm ~/.python27_compiled/*
  2. ipcluster start and then run correctness_tests/parallel-test.py (maybe from the console, just to cut out any notebook weirdness)

Let me know what happens.

mattjj commented 11 years ago

Running now!