I am working on recreating the baseline solutions. I have completed the first 7 steps. When I run "python make_submission.py mlp.pkl public_test" I get the following error\r\n while loading the pkl file:
\r\n File "C:\\...\\Theano\\theano\\sandbox\\cuda\\type.py", line 469, in CudaNdarray_unpickler
\r\n return cuda.CudaNdarray(npa)
\r\nAttributeError: ("'NoneType' object has no attribute 'CudaNdarray'", <function CudaNdarray_unpickler ...
This is the first time I've used Theano. Is there any additional configuration of Theano I need to do to get this to run?
\r\nThanks,
\r\nSteve
", "messages": [ { "id": "23373", "postDate": "04/23/2013 18:35:18", "content": "I am working on recreating the baseline solutions. I have completed the first 7 steps. When I run "python make_submission.py mlp.pkl public_test" I get the following error\r\n while loading the pkl file:
\r\n File "C:\\...\\Theano\\theano\\sandbox\\cuda\\type.py", line 469, in CudaNdarray_unpickler
\r\n return cuda.CudaNdarray(npa)
\r\nAttributeError: ("'NoneType' object has no attribute 'CudaNdarray'", <function CudaNdarray_unpickler ...
This is the first time I've used Theano. Is there any additional configuration of Theano I need to do to get this to run?
\r\nThanks,
\r\nSteve
", "rawMarkdown": "", "votes": null }, { "id": "23376", "postDate": "04/23/2013 19:02:03", "content": "If you trained the model on a GPU, you'll need to unpickle on a computer that has a GPU available too. Do you think that's what's going on?
", "rawMarkdown": "", "votes": null }, { "id": "23378", "postDate": "04/23/2013 19:39:10", "content": "My graphics card is an Intel Graphics Media Accelerator HD. GPU is Arrandale.
\r\nDoes it need to be a CUDA-enabled GPU to load the mlp.pkl model?
\r\nThanks,
\r\nSteve
", "rawMarkdown": "", "votes": null }, { "id": "23380", "postDate": "04/23/2013 19:43:02", "content": "Who made the .pkl file, on which machine? Did you train it yourself, on that same machine? If so, it's weird that theano is trying to use the CudaNdarrayUnpickler.
", "rawMarkdown": "", "votes": null }, { "id": "23381", "postDate": "04/23/2013 19:48:12", "content": "At this point I am just trying to load the mlp.pkl file that I downloaded from http://www-etud.iro.umontreal.ca/~goodfeli/mlp.pkl in step #7 of the "run\r\n the example baselines" instructions. Thank you.
", "rawMarkdown": "", "votes": null }, { "id": "23382", "postDate": "04/23/2013 19:52:22", "content": "Oh, sorry. I thought I had converted everything back to numpy ndarrays, but I guess not. I'll post here again when I've updated the pkl file to work without GPU. I really hate the way theano pickles things.
", "rawMarkdown": "", "votes": null }, { "id": "23385", "postDate": "04/23/2013 21:32:43", "content": "I'm sorry. Theano is doing something that I just don't understand, and I have no idea how to fix it. My conversion script is evidently missing a CUDA variable that theano has created behind the scenes. If I try to step through the pickle file manually and\r\n look for all possible theano variables, something causes a seg fault and crashes the python interpreter, probably indicating a bug in Theano's C code somewhere. If you can get access to a CUDA-enabled GPU, the pickle files should work. I'll write to the theano-users\r\n e-mail list to see if anyone knows how to convert it, but Frederic Bastien says there's no easy way, and he's the person who's most likely to know how.
", "rawMarkdown": "", "votes": null }, { "id": "23391", "postDate": "04/24/2013 00:06:16", "content": "Okay. Thank you for looking into it.
", "rawMarkdown": "", "votes": null }, { "id": "23394", "postDate": "04/24/2013 01:30:07", "content": "I think I've fixed it! I managed to parse the object with different calls that luckily happened to avoid whatever made theano seg fault. Try using this file:
\r\nhttp://www-etud.iro.umontreal.ca/~goodfeli/logreg_best_cpu.pkl
\r\nIf it works for you I'll update the README in the main repository.
", "rawMarkdown": "", "votes": null }, { "id": "23395", "postDate": "04/24/2013 01:34:46", "content": "Luckily, the same script also worked for the mlp:
\r\nhttp://www-etud.iro.umontreal.ca/~goodfeli/mlp_cpu.pkl
", "rawMarkdown": "", "votes": null }, { "id": "23397", "postDate": "04/24/2013 02:39:00", "content": "That seems to have done the trick! I ran both models for a subset of the images, and they both worked. And I assume it will scale up to the full 500 images without any problem.
\r\nThanks again!
", "rawMarkdown": "", "votes": null } ], "comments": [ { "id": 23376, "author_name": "iangoodfellow", "author_url": "", "post_date": "04/23/2013 19:02:03", "content": "If you trained the model on a GPU, you'll need to unpickle on a computer that has a GPU available too. Do you think that's what's going on?
", "votes": null, "replies": [] }, { "id": 23378, "author_name": "breakfastpirate", "author_url": "", "post_date": "04/23/2013 19:39:10", "content": "My graphics card is an Intel Graphics Media Accelerator HD. GPU is Arrandale.
\r\nDoes it need to be a CUDA-enabled GPU to load the mlp.pkl model?
\r\nThanks,
\r\nSteve
", "votes": null, "replies": [] }, { "id": 23380, "author_name": "iangoodfellow", "author_url": "", "post_date": "04/23/2013 19:43:02", "content": "Who made the .pkl file, on which machine? Did you train it yourself, on that same machine? If so, it's weird that theano is trying to use the CudaNdarrayUnpickler.
", "votes": null, "replies": [] }, { "id": 23381, "author_name": "breakfastpirate", "author_url": "", "post_date": "04/23/2013 19:48:12", "content": "At this point I am just trying to load the mlp.pkl file that I downloaded from http://www-etud.iro.umontreal.ca/~goodfeli/mlp.pkl in step #7 of the "run\r\n the example baselines" instructions. Thank you.
", "votes": null, "replies": [] }, { "id": 23382, "author_name": "iangoodfellow", "author_url": "", "post_date": "04/23/2013 19:52:22", "content": "Oh, sorry. I thought I had converted everything back to numpy ndarrays, but I guess not. I'll post here again when I've updated the pkl file to work without GPU. I really hate the way theano pickles things.
", "votes": null, "replies": [] }, { "id": 23385, "author_name": "iangoodfellow", "author_url": "", "post_date": "04/23/2013 21:32:43", "content": "I'm sorry. Theano is doing something that I just don't understand, and I have no idea how to fix it. My conversion script is evidently missing a CUDA variable that theano has created behind the scenes. If I try to step through the pickle file manually and\r\n look for all possible theano variables, something causes a seg fault and crashes the python interpreter, probably indicating a bug in Theano's C code somewhere. If you can get access to a CUDA-enabled GPU, the pickle files should work. I'll write to the theano-users\r\n e-mail list to see if anyone knows how to convert it, but Frederic Bastien says there's no easy way, and he's the person who's most likely to know how.
", "votes": null, "replies": [] }, { "id": 23391, "author_name": "breakfastpirate", "author_url": "", "post_date": "04/24/2013 00:06:16", "content": "Okay. Thank you for looking into it.
", "votes": null, "replies": [] }, { "id": 23394, "author_name": "iangoodfellow", "author_url": "", "post_date": "04/24/2013 01:30:07", "content": "I think I've fixed it! I managed to parse the object with different calls that luckily happened to avoid whatever made theano seg fault. Try using this file:
\r\nhttp://www-etud.iro.umontreal.ca/~goodfeli/logreg_best_cpu.pkl
\r\nIf it works for you I'll update the README in the main repository.
", "votes": null, "replies": [] }, { "id": 23395, "author_name": "iangoodfellow", "author_url": "", "post_date": "04/24/2013 01:34:46", "content": "Luckily, the same script also worked for the mlp:
\r\nhttp://www-etud.iro.umontreal.ca/~goodfeli/mlp_cpu.pkl
", "votes": null, "replies": [] }, { "id": 23397, "author_name": "breakfastpirate", "author_url": "", "post_date": "04/24/2013 02:39:00", "content": "That seems to have done the trick! I ran both models for a subset of the images, and they both worked. And I assume it will scale up to the full 500 images without any problem.
\r\nThanks again!
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "23373": "", "23376": "", "23378": "", "23380": "", "23381": "", "23382": "", "23385": "", "23391": "", "23394": "", "23395": "", "23397": "" }, "source": "meta" }