Skip to content
Snippets Groups Projects
Commit d8c97cdd authored by Neta Zmora's avatar Neta Zmora
Browse files

AMC: Revive support for weights-channels removal

This is in contrast to weights-filters removal
parent 09d2eea3
No related branches found
No related tags found
No related merge requests found
...@@ -226,14 +226,14 @@ def do_adc_internal(model, args, optimizer_data, validate_fn, save_checkpoint_fn ...@@ -226,14 +226,14 @@ def do_adc_internal(model, args, optimizer_data, validate_fn, save_checkpoint_fn
# Create a dictionary of parameters that Coach will handover to DistillerWrapperEnvironment # Create a dictionary of parameters that Coach will handover to DistillerWrapperEnvironment
# Once it creates it. # Once it creates it.
services = distiller.utils.MutableNamedTuple({ services = distiller.utils.MutableNamedTuple({
'validate_fn': validate_fn, 'validate_fn': validate_fn,
'save_checkpoint_fn': save_checkpoint_fn, 'save_checkpoint_fn': save_checkpoint_fn,
'train_fn': train_fn}) 'train_fn': train_fn})
app_args = distiller.utils.MutableNamedTuple({ app_args = distiller.utils.MutableNamedTuple({
'dataset': dataset, 'dataset': dataset,
'arch': arch, 'arch': arch,
'optimizer_data': optimizer_data}) 'optimizer_data': optimizer_data})
amc_cfg = distiller.utils.MutableNamedTuple({ amc_cfg = distiller.utils.MutableNamedTuple({
'protocol': args.amc_protocol, 'protocol': args.amc_protocol,
...@@ -243,7 +243,8 @@ def do_adc_internal(model, args, optimizer_data, validate_fn, save_checkpoint_fn ...@@ -243,7 +243,8 @@ def do_adc_internal(model, args, optimizer_data, validate_fn, save_checkpoint_fn
'action_range': action_range, 'action_range': action_range,
'conv_cnt': conv_cnt, 'conv_cnt': conv_cnt,
'reward_frequency': args.amc_reward_frequency, 'reward_frequency': args.amc_reward_frequency,
'ft_frequency': args.amc_ft_frequency}) 'ft_frequency': args.amc_ft_frequency,
'pruning_pattern': "filters"}) # "channels"}) #
#net_wrapper = NetworkWrapper(model, app_args, services) #net_wrapper = NetworkWrapper(model, app_args, services)
#return sample_networks(net_wrapper, services) #return sample_networks(net_wrapper, services)
...@@ -344,9 +345,21 @@ resnet56_params = ["module.layer1.0.conv1.weight", "module.layer1.1.conv1.weight ...@@ -344,9 +345,21 @@ resnet56_params = ["module.layer1.0.conv1.weight", "module.layer1.1.conv1.weight
"module.layer3.3.conv1.weight", "module.layer3.4.conv1.weight", "module.layer3.5.conv1.weight", "module.layer3.3.conv1.weight", "module.layer3.4.conv1.weight", "module.layer3.5.conv1.weight",
"module.layer3.6.conv1.weight", "module.layer3.7.conv1.weight", "module.layer3.8.conv1.weight"] "module.layer3.6.conv1.weight", "module.layer3.7.conv1.weight", "module.layer3.8.conv1.weight"]
plain20_params = ["module.layer1.0.conv1.weight", "module.layer1.0.conv2.weight",
"module.layer1.1.conv1.weight", "module.layer1.1.conv2.weight",
"module.layer1.2.conv1.weight", "module.layer1.2.conv2.weight",
"module.layer2.0.conv1.weight", "module.layer2.0.conv2.weight",
"module.layer2.1.conv1.weight", "module.layer2.1.conv2.weight",
"module.layer2.2.conv1.weight", "module.layer2.2.conv2.weight",
"module.layer3.0.conv1.weight", "module.layer3.0.conv2.weight",
"module.layer3.1.conv1.weight", "module.layer3.1.conv2.weight",
"module.layer3.2.conv1.weight", "module.layer3.2.conv2.weight"]
resnet50_layers = [param[:-len(".weight")] for param in resnet50_params] resnet50_layers = [param[:-len(".weight")] for param in resnet50_params]
resnet20_layers = [param[:-len(".weight")] for param in resnet20_params] resnet20_layers = [param[:-len(".weight")] for param in resnet20_params]
resnet56_layers = [param[:-len(".weight")] for param in resnet56_params] resnet56_layers = [param[:-len(".weight")] for param in resnet56_params]
plain20_layers = [param[:-len(".weight")] for param in plain20_params]
class NetworkWrapper(object): class NetworkWrapper(object):
...@@ -375,6 +388,8 @@ class NetworkWrapper(object): ...@@ -375,6 +388,8 @@ class NetworkWrapper(object):
resnet_layers = resnet56_layers resnet_layers = resnet56_layers
elif self.app_args.arch == "resnet50": elif self.app_args.arch == "resnet50":
resnet_layers = resnet50_layers resnet_layers = resnet50_layers
elif self.app_args.arch == "plain20_cifar":
resnet_layers = plain20_layers
return collect_conv_details(model, self.app_args.dataset, True, resnet_layers) return collect_conv_details(model, self.app_args.dataset, True, resnet_layers)
def num_layers(self): def num_layers(self):
...@@ -426,16 +441,16 @@ class NetworkWrapper(object): ...@@ -426,16 +441,16 @@ class NetworkWrapper(object):
conv_pname = layer.name + ".weight" conv_pname = layer.name + ".weight"
conv_p = distiller.model_find_param(self.model, conv_pname) conv_p = distiller.model_find_param(self.model, conv_pname)
msglogger.info("ADC: removing %.1f%% %s from %s" % (fraction_to_prune*100, prune_what, conv_pname)) msglogger.info("ADC: trying to remove %.1f%% %s from %s" % (fraction_to_prune*100, prune_what, conv_pname))
if prune_what == "channels": if prune_what == "channels":
calculate_sparsity = distiller.sparsity_ch calculate_sparsity = distiller.sparsity_ch
remove_structures = distiller.remove_channels remove_structures_fn = distiller.remove_channels
group_type = "Channels" group_type = "Channels"
elif prune_what == "filters": elif prune_what == "filters":
calculate_sparsity = distiller.sparsity_3D calculate_sparsity = distiller.sparsity_3D
group_type = "Filters" group_type = "Filters"
remove_structures = distiller.remove_filters remove_structures_fn = distiller.remove_filters
else: else:
raise ValueError("unsupported structure {}".format(prune_what)) raise ValueError("unsupported structure {}".format(prune_what))
# Create a channel-ranking pruner # Create a channel-ranking pruner
...@@ -446,12 +461,12 @@ class NetworkWrapper(object): ...@@ -446,12 +461,12 @@ class NetworkWrapper(object):
if (self.zeros_mask_dict[conv_pname].mask is None or if (self.zeros_mask_dict[conv_pname].mask is None or
calculate_sparsity(self.zeros_mask_dict[conv_pname].mask) == 0): calculate_sparsity(self.zeros_mask_dict[conv_pname].mask) == 0):
msglogger.info("remove_structures: aborting because there are no channels to prune") msglogger.info("remove_structures: aborting because there are no structures to prune")
return 0 return 0
# Use the mask to prune # Use the mask to prune
self.zeros_mask_dict[conv_pname].apply_mask(conv_p) self.zeros_mask_dict[conv_pname].apply_mask(conv_p)
remove_structures(self.model, self.zeros_mask_dict, self.app_args.arch, self.app_args.dataset, optimizer=None) remove_structures_fn(self.model, self.zeros_mask_dict, self.app_args.arch, self.app_args.dataset, optimizer=None)
conv_p = distiller.model_find_param(self.model, conv_pname) conv_p = distiller.model_find_param(self.model, conv_pname)
return 1 - (self.get_layer_macs(layer) / macs_before) return 1 - (self.get_layer_macs(layer) / macs_before)
...@@ -476,8 +491,6 @@ class NetworkWrapper(object): ...@@ -476,8 +491,6 @@ class NetworkWrapper(object):
class DistillerWrapperEnvironment(gym.Env): class DistillerWrapperEnvironment(gym.Env):
metadata = {'render.modes': ['human']}
def __init__(self, model, app_args, amc_cfg, services): def __init__(self, model, app_args, amc_cfg, services):
self.pylogger = distiller.data_loggers.PythonLogger(msglogger) self.pylogger = distiller.data_loggers.PythonLogger(msglogger)
self.tflogger = distiller.data_loggers.TensorBoardLogger(msglogger.logdir) self.tflogger = distiller.data_loggers.TensorBoardLogger(msglogger.logdir)
...@@ -573,23 +586,26 @@ class DistillerWrapperEnvironment(gym.Env): ...@@ -573,23 +586,26 @@ class DistillerWrapperEnvironment(gym.Env):
def step(self, pruning_action): def step(self, pruning_action):
"""Take a step, given an action. """Take a step, given an action.
The action represents the desired sparsity. The action represents the desired sparsity for the "current" layer.
This function is invoked by the Agent. This function is invoked by the Agent.
""" """
msglogger.info("env.step - current_layer_id={} episode={}".format(self.current_layer_id, self.episode)) msglogger.info("env.step - current_layer_id={} episode={}".format(self.current_layer_id, self.episode))
pruning_action = pruning_action[0]
msglogger.info("\tAgent pruning_action={}".format(pruning_action)) msglogger.info("\tAgent pruning_action={}".format(pruning_action))
self.agent_action_history.append(pruning_action)
if is_using_continuous_action_space(self.amc_cfg.agent_algo): if is_using_continuous_action_space(self.amc_cfg.agent_algo):
pruning_action = np.clip(pruning_action[0], self.action_low, self.action_high) pruning_action = np.clip(pruning_action, self.action_low, self.action_high)
else: else:
# Divide the action space into 10 discrete levels (0%, 10%, 20%,....90% sparsity) # Divide the action space into 10 discrete levels (0%, 10%, 20%,....90% sparsity)
pruning_action = pruning_action / 10 pruning_action = pruning_action / 10
msglogger.info("\tAgent clipped pruning_action={}".format(pruning_action)) msglogger.info("\tAgent clipped pruning_action={}".format(pruning_action))
self.agent_action_history.append(pruning_action)
if self.amc_cfg.action_constrain_fn is not None: if self.amc_cfg.action_constrain_fn is not None:
pruning_action = self.amc_cfg.action_constrain_fn(self, pruning_action=pruning_action) pruning_action = self.amc_cfg.action_constrain_fn(self, pruning_action=pruning_action)
msglogger.info("Constrained pruning_action={}".format(pruning_action)) msglogger.info("Constrained pruning_action={}".format(pruning_action))
# Calculate the final compression rate
total_macs_before, _ = self.net_wrapper.get_model_resources_requirements(self.model) total_macs_before, _ = self.net_wrapper.get_model_resources_requirements(self.model)
layer_macs = self.net_wrapper.get_layer_macs(self.current_layer()) layer_macs = self.net_wrapper.get_layer_macs(self.current_layer())
msglogger.info("\tlayer_macs={:.2f}".format(layer_macs / self.dense_model_macs)) msglogger.info("\tlayer_macs={:.2f}".format(layer_macs / self.dense_model_macs))
...@@ -599,7 +615,7 @@ class DistillerWrapperEnvironment(gym.Env): ...@@ -599,7 +615,7 @@ class DistillerWrapperEnvironment(gym.Env):
if pruning_action > 0: if pruning_action > 0:
pruning_action = self.net_wrapper.remove_structures(self.current_layer_id, pruning_action = self.net_wrapper.remove_structures(self.current_layer_id,
fraction_to_prune=pruning_action, fraction_to_prune=pruning_action,
prune_what="filters") prune_what=self.amc_cfg.pruning_pattern)
else: else:
pruning_action = 0 pruning_action = 0
......
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment