state.nepochs = [128] # Different validation runs # - 100 training examples (x20 different samples of 100 training examples) # - 1000 training examples (x10 different samples of 1000 training examples) # - 10000 training examples (x1 different sample of 10000 training examples) # (because of jobman, the keys have to be strings, not ints) # NOTE: Probably you don't want to make trainsize larger than 10K, # because it will be too large for CPU memory. state.validation_runs_for_each_trainingsize = {"100": 20, "1000": 10, "10000": 1} # For each layer, a list of the epochs at which you evaluate the # reconstruction error and linear-SVM-supervised error. # All the different results you have from here will be stored in a # separate file per layer. state.epochstest = [[0,2,4,8,12,16,24,32,48,64,96,128]] #epochstest = [[0,5,30],[0,5,30],[0,2,4,8,16,30]] state.BATCH_TEST = 100 state.BATCH_CREATION_LIBSVM = 500 state.NB_MAX_TRAINING_EXAMPLES_SVM = 10000 #NB_MAX_TRAINING_EXAMPLES_SVM = 1000 # FIXME: Change back to 10000 <======================================================================== # 1000 is just for fast running during development #NB_MAX_TRAINING_EXAMPLES_SVM = 100 # FIXME: Change back to 10000 <======================================================================== # # 100 is just for superfast running during development state.SVM_INITIALC = 0.001 state.SVM_STEPFACTOR = 10. state.SVM_MAXSTEPS = 10 #hardcoded path to your liblinear source:
# - 1000 training examples (x10 different samples of 1000 training examples) # - 10000 training examples (x1 different sample of 10000 training examples) # (because of jobman, the keys have to be strings, not ints) # NOTE: Probably you don't want to make trainsize larger than 10K, # because it will be too large for CPU memory. state.validation_runs_for_each_trainingsize = { "100": 20, "1000": 10, "10000": 1 } # For each layer, a list of the epochs at which you evaluate the # reconstruction error and linear-SVM-supervised error. # All the different results you have from here will be stored in a # separate file per layer. state.epochstest = [[0, 1, 2, 3, 4, 6, 8, 11, 15, 30]] #epochstest = [[0,5,30],[0,5,30],[0,2,4,8,16,30]] state.BATCH_TEST = 100 state.BATCH_CREATION_LIBSVM = 500 state.NB_MAX_TRAINING_EXAMPLES_SVM = 10000 #NB_MAX_TRAINING_EXAMPLES_SVM = 1000 # FIXME: Change back to 10000 <======================================================================== # 1000 is just for fast running during development #NB_MAX_TRAINING_EXAMPLES_SVM = 100 # FIXME: Change back to 10000 <======================================================================== # # 100 is just for superfast running during development state.SVM_INITIALC = 0.001 state.SVM_STEPFACTOR = 10. state.SVM_MAXSTEPS = 10 #hardcoded path to your liblinear source:
state.nepochs = [30] # Different validation runs # - 100 training examples (x20 different samples of 100 training examples) # - 1000 training examples (x10 different samples of 1000 training examples) # - 10000 training examples (x1 different sample of 10000 training examples) # (because of jobman, the keys have to be strings, not ints) # NOTE: Probably you don't want to make trainsize larger than 10K, # because it will be too large for CPU memory. state.validation_runs_for_each_trainingsize = {"100": 20, "1000": 10, "10000": 1} # For each layer, a list of the epochs at which you evaluate the # reconstruction error and linear-SVM-supervised error. # All the different results you have from here will be stored in a # separate file per layer. state.epochstest = [[0,1,2,3,4,6,8,11,15,30]] #epochstest = [[0,5,30],[0,5,30],[0,2,4,8,16,30]] state.BATCH_TEST = 100 state.BATCH_CREATION_LIBSVM = 500 state.NB_MAX_TRAINING_EXAMPLES_SVM = 10000 #NB_MAX_TRAINING_EXAMPLES_SVM = 1000 # FIXME: Change back to 10000 <======================================================================== # 1000 is just for fast running during development #NB_MAX_TRAINING_EXAMPLES_SVM = 100 # FIXME: Change back to 10000 <======================================================================== # # 100 is just for superfast running during development state.SVM_INITIALC = 0.001 state.SVM_STEPFACTOR = 10. state.SVM_MAXSTEPS = 10 #hardcoded path to your liblinear source: