akolishchak / doom-net-pytorch Goto Github PK

I am trying to run the imitation learning code in this repo, but "test/datasets/vizdoom/cig_map02/flat.h5" is not in this repo. Could you please share the demonstration data you used when running behavior cloning.

--- Thanks Kianté

Hi,

Just wanted know to know if you were able to train Behavioral cloning + LSTM model successfully? For me, I am not able to train it.

Hi,
Impressive work!
I am new to this，got an error when run the example by commands resume/test， training command works fine.

after training, got the pth file: d3_battle__ppo_cp.pth_optimizer.pth_7000.pth
I edited the ckpt path in d3_ppo.sh:
CHECK_POINT=$BASEDIR/checkpoints/d3_battle__ppo_cp.pth_optimizer.pth_7000.pth

the commands with probrom i use is:
sudo ./d3_ppo.sh test
sudo ./d3_ppo.sh resume

error message:
Traceback (most recent call last):
File "/media/shar/Runtime/LinRuntime/doom-net/src/main.py", line 43, in
model = get_model(args)
File "/media/shar/Runtime/LinRuntime/doom-net/src/model_utils.py", line 52, in get_model
model = model_classargs.model
File "/media/shar/Runtime/LinRuntime/doom-net/src/ppo.py", line 151, in init
self.model.load_state_dict(state_dict)
File "/usr/local/lib/python3.5/dist-packages/torch/nn/modules/module.py", line 777, in load_state_dict
self.class.name, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for BaseModel:
Missing key(s) in state_dict: "conv1.weight", "conv1.bias", "conv2.weight", "conv2.bias", "conv3.weight", "conv3.bias", "conv4.weight", "conv4.bias", "conv5.weight", "conv5.bias", "conv6.weight", "conv6.bias", "screen_features1.bias_hh", "screen_features1.weight_hh", "screen_features1.weight_ih", "screen_features1.bias_ih", "batch_norm.running_mean", "batch_norm.running_var", "batch_norm.weight", "batch_norm.bias", "action1.weight", "action1.bias", "action2.weight", "action2.bias", "value1.weight", "value1.bias", "value2.weight", "value2.bias".
Unexpected key(s) in state_dict: "state", "param_groups".

any advice how can i run it right?
thanks!!

where is the doom_instance_map.py file, can't find this file

This is awesome!
I am new to all of this but trying to learn as i go, so far i have vizdoom + dependencies installed and the theano, pytorch and tensorflow examples (that came included with vizdoom) all work.

sorry for such a basic question, but how do i install this into vizdoom?

Thank you very much, any help would be greatly appreciated.

Hi,
Within "ppo net" in the code, when i set the --batch_size ==10, it works fine, but then i set it to 1, the network learned nothing after 10 times episodes, could you give some idea?

Thanks

for example, what is action_set_speed_shot_backward_right meaning? thank you.