Error: PolicyBasedRL

[NotSure2732]

Describe the issue: I tried running the the following models space with PolicyBasedRL and I will also put in the experiment configuration:

BASELINE NAS USING v2.7

from nni.retiarii.serializer import model_wrapper import torch.nn.functional as F import nni.retiarii.nn.pytorch as nn

class Block1(nn.Module): def init(self, layer_size): super().init() self.conv1 = nn.Conv2d(3, layer_size2, 3, stride=1,padding=1) self.pool = nn.MaxPool2d(2, 2) self.conv2 = nn.Conv2d(layer_size2, layer_size*8, 3, stride=1, padding=1)

def forward(self, x):
    x = F.relu(self.conv1(x))
    x = self.pool(x)
    x = F.relu(self.conv2(x))
    x = self.pool(x)
    return x

class Block2(nn.Module): def init(self, layer_size): super().init() self.conv1 = nn.Conv2d(3, layer_size, 3, stride=1,padding=1) self.conv2 = nn.Conv2d(layer_size, layer_size2, 3, stride=1,padding=1) self.pool = nn.MaxPool2d(2, 2) self.conv3 = nn.Conv2d(layer_size2, layer_size*8, 3, stride=1,padding=1)

def forward(self, x):
    x = F.relu(self.conv1(x))
    x = F.relu(self.conv2(x))
    x = self.pool(x)
    x = F.relu(self.conv3(x))
    x = self.pool(x)

    return x

class Block3(nn.Module): def init(self, layer_size): super().init() self.conv1 = nn.Conv2d(3, layer_size, 3, stride=1,padding=1) self.conv2 = nn.Conv2d(layer_size, layer_size2, 3, stride=1,padding=1) self.pool = nn.MaxPool2d(2, 2) self.conv3 = nn.Conv2d(layer_size2, layer_size4, 3, stride=1,padding=1) self.conv4 = nn.Conv2d(layer_size4, layer_size*8, 3, stride=1, padding=1)

def forward(self, x):
    x = F.relu(self.conv1(x))
    x = F.relu(self.conv2(x))
    x = self.pool(x)
    x = F.relu(self.conv3(x))
    x = F.relu(self.conv4(x))
    x = self.pool(x)

    return x

@model_wrapper class Net(nn.Module): def init(self): super().init() rand_var = nn.ValueChoice([32,64]) self.conv1 = nn.LayerChoice([Block1(rand_var),Block2(rand_var),Block3(rand_var)]) self.conv2 = nn.Conv2d(rand_var8,rand_var16 , 3, stride=1, padding=1) self.fc1 = nn.Linear(rand_var168*8, 120) self.fc2 = nn.Linear(120, 84) self.fc3 = nn.Linear(84, 10)

def forward(self, x):

    x = self.conv1(x)
    x = F.relu(self.conv2(x))   
    x = x.reshape(x.shape[0],-1)
    x = F.relu(self.fc1(x))
    x = F.relu(self.fc2(x))
    x = self.fc3(x)
    return x

model = Net()

from nni.retiarii.experiment.pytorch import RetiariiExeConfig, RetiariiExperiment

exp = RetiariiExperiment(model, trainer, [], RL_strategy)

exp_config = RetiariiExeConfig('local') exp_config.experiment_name = '5%_RL_10_epochs_64_batch' exp_config.trial_concurrency = 2 exp_config.max_trial_number = 100

exp_config.trial_gpu_number = 2

exp_config.max_experiment_duration = '660m' exp_config.execution_engine = 'base' exp_config.training_service.use_active_gpu = False

--> This led to the following error:

[2022-04-24 23:49:22] ERROR (nni.runtime.msg_dispatcher_base/Thread-5) 3 Traceback (most recent call last): File "/Users/sh/opt/anaconda3/lib/python3.7/site-packages/nni/runtime/msg_dispatcher_base.py", line 88, in command_queue_worker self.process_command(command, data) File "/Users/sh/opt/anaconda3/lib/python3.7/site-packages/nni/runtime/msg_dispatcher_base.py", line 147, in process_command command_handlerscommand File "/Users/sh/opt/anaconda3/lib/python3.7/site-packages/nni/retiarii/integration.py", line 170, in handle_report_metric_data self._process_value(data['value'])) File "/Users/sh/opt/anaconda3/lib/python3.7/site-packages/nni/retiarii/execution/base.py", line 111, in _intermediate_metric_callback model = self._running_models[trial_id] KeyError: 3

What does this error mean/ why does it occur/ how can I fix it?

Thanks for your help!

Environment:

• NNI version:
• Training service (local|remote|pai|aml|etc):
• Client OS:
• Server OS (for remote mode only):
• Python version:
• PyTorch/TensorFlow version:
• Is conda/virtualenv/venv used?:
• Is running in Docker?:

Configuration:

• Experiment config (remember to remove secrets!):
• Search space:

Log message:

• nnimanager.log:
• dispatcher.log:
• nnictl stdout and stderr:

How to reproduce it?:

[matluster]

Let's narrow down the cause of your issue.

• You are using base execution engine. Did you experience the same problem with python execution engine?
• You are using RL strategy. Did you experience the same problem with Evolution or random strategy?

One more suggestion here: please format your issue description with markdown syntax properly. Otherwise, your code is not human-readable.

[GoWithWind2015]

this problem is resolve in v2.10 : "Fixed the racing issue in RL strategy on submitting models."