Mercurial > ift6266
comparison deep/convolutional_dae/salah_exp/utils.py @ 358:31641a84e0ae
Initial commit for the experimental setup of the denoising convolutional network
| author | humel |
|---|---|
| date | Thu, 22 Apr 2010 00:49:42 -0400 |
| parents | |
| children |
comparison
equal
deleted
inserted
replaced
| 357:9a7b74927f7d | 358:31641a84e0ae |
|---|---|
| 1 #!/usr/bin/python | |
| 2 # coding: utf-8 | |
| 3 | |
| 4 from __future__ import with_statement | |
| 5 | |
| 6 from jobman import DD | |
| 7 | |
| 8 # from pylearn codebase | |
| 9 # useful in __init__(param1, param2, etc.) to save | |
| 10 # values in self.param1, self.param2... just call | |
| 11 # update_locals(self, locals()) | |
| 12 def update_locals(obj, dct): | |
| 13 if 'self' in dct: | |
| 14 del dct['self'] | |
| 15 obj.__dict__.update(dct) | |
| 16 | |
| 17 # from a dictionary of possible values for hyperparameters, e.g. | |
| 18 # hp_values = {'learning_rate':[0.1, 0.01], 'num_layers': [1,2]} | |
| 19 # create a list of other dictionaries representing all the possible | |
| 20 # combinations, thus in this example creating: | |
| 21 # [{'learning_rate': 0.1, 'num_layers': 1}, ...] | |
| 22 # (similarly for combinations (0.1, 2), (0.01, 1), (0.01, 2)) | |
| 23 def produit_cartesien_jobs(val_dict): | |
| 24 job_list = [DD()] | |
| 25 all_keys = val_dict.keys() | |
| 26 | |
| 27 for key in all_keys: | |
| 28 possible_values = val_dict[key] | |
| 29 new_job_list = [] | |
| 30 for val in possible_values: | |
| 31 for job in job_list: | |
| 32 to_insert = job.copy() | |
| 33 to_insert.update({key: val}) | |
| 34 new_job_list.append(to_insert) | |
| 35 job_list = new_job_list | |
| 36 | |
| 37 return job_list | |
| 38 | |
| 39 def test_produit_cartesien_jobs(): | |
| 40 vals = {'a': [1,2], 'b': [3,4,5]} | |
| 41 print produit_cartesien_jobs(vals) | |
| 42 | |
| 43 | |
| 44 # taken from http://stackoverflow.com/questions/276052/how-to-get-current-cpu-and-ram-usage-in-python | |
| 45 """Simple module for getting amount of memory used by a specified user's | |
| 46 processes on a UNIX system. | |
| 47 It uses UNIX ps utility to get the memory usage for a specified username and | |
| 48 pipe it to awk for summing up per application memory usage and return the total. | |
| 49 Python's Popen() from subprocess module is used for spawning ps and awk. | |
| 50 | |
| 51 """ | |
| 52 | |
| 53 import subprocess | |
| 54 | |
| 55 class MemoryMonitor(object): | |
| 56 | |
| 57 def __init__(self, username): | |
| 58 """Create new MemoryMonitor instance.""" | |
| 59 self.username = username | |
| 60 | |
| 61 def usage(self): | |
| 62 """Return int containing memory used by user's processes.""" | |
| 63 self.process = subprocess.Popen("ps -u %s -o rss | awk '{sum+=$1} END {print sum}'" % self.username, | |
| 64 shell=True, | |
| 65 stdout=subprocess.PIPE, | |
| 66 ) | |
| 67 self.stdout_list = self.process.communicate()[0].split('\n') | |
| 68 return int(self.stdout_list[0]) | |
| 69 |