Hi I am trying to do my first unity ml-agents ai. Before, when I wanted to train my ai, I was writing
mlagents-learn config/trainer_config.yaml --run-id=Taxi-1 --train
in the terminal, but the ai stopped training after 50 000 steps. Then, I tried to train it again, with another
mlagents-learn config/trainer_config.yaml --run-id=Taxi-1 --train
Then, I saw that you had to add --load to the command if you wanted it not to start over the whole training and to continue training your previous model. However, when I write
mlagents-learn config/trainer_config.yaml --load --run-id=Taxi-1 --train
It only does one step and then stops. This is what it writes in the terminal:
INFO:mlagents.trainers:{'--curriculum': 'None',
'--docker-target-name': 'None',
'--env': 'None',
'--help': False,
'--keep-checkpoints': '5',
'--lesson': '0',
'--load': True,
'--no-graphics': False,
'--num-runs': '1',
'--run-id': 'Taxi-1',
'--save-freq': '50000',
'--seed': '-1',
'--slow': False,
'--train': True,
'--worker-id': '0',
'<trainer-config-path>': 'config/trainer_config.yaml'}
INFO:mlagents.envs:Start training by pressing the Play button in the
Unity Editor.
INFO:mlagents.envs:
'Academy' started successfully!
Unity Academy name: Academy
Number of Brains: 2
Number of Training Brains : 1
Reset Parameters :
Unity brain name: CarLBrain
Number of Visual Observations (per agent): 0
Vector Observation space size (per agent): 12
Number of stacked Vector Observation: 6
Vector Action space type: continuous
Vector Action space size (per agent): [2]
Vector Action descriptions: ,
Unity brain name: CarPBrain
Number of Visual Observations (per agent): 0
Vector Observation space size (per agent): 12
Number of stacked Vector Observation: 6
Vector Action space type: discrete
Vector Action space size (per agent): [10, 10]
Vector Action descriptions: ,
INFO:mlagents.trainers:Loading Model for brain CarLBrain
INFO:tensorflow:Restoring parameters from ./models/Taxi-1-
0/CarLBrain/model-50001.cptk
INFO:mlagents.envs:Hyperparameters for the PPO Trainer of brain
CarLBrain:
batch_size: 1024
beta: 0.005
buffer_size: 10240
epsilon: 0.2
gamma: 0.99
hidden_units: 128
lambd: 0.95
learning_rate: 0.0003
max_steps: 5.0e4
normalize: False
num_epoch: 3
num_layers: 2
time_horizon: 64
sequence_length: 64
summary_freq: 1000
use_recurrent: False
summary_path: ./summaries/Taxi-1-0_CarLBrain
memory_size: 256
use_curiosity: False
curiosity_strength: 0.01
curiosity_enc_size: 128
model_path: ./models/Taxi-1-0/CarLBrain
INFO:mlagents.envs:Saved Model
INFO:mlagents.trainers:List of nodes to export for brain :CarLBrain
INFO:mlagents.trainers: is_continuous_control
INFO:mlagents.trainers: version_number
INFO:mlagents.trainers: memory_size
INFO:mlagents.trainers: action_output_shape
INFO:mlagents.trainers: action
INFO:mlagents.trainers: action_probs
INFO:mlagents.trainers: value_estimate
INFO:tensorflow:Restoring parameters from ./models/Taxi-1-
0/CarLBrain/model-50002.cptk
INFO:tensorflow:Froze 17 variables.
Converted 17 variables to const ops.
Do you know how I can continue my training to more than 50 000 steps? Thank you for your help! Don't hesitate to ask for any clarification.
