isl-org / directfutureprediction Goto Github PK

I tried to run the D3_battle example but got following error about "list index out of range"

thinh@thinh-H67MU3:~/drl/DirectFuturePrediction/examples/D3_battle$ python3 run_exp.py /usr/local/lib/python3.5/dist-packages/vizdoom/vizdoom.so Requested resolution not supported: <class 'AttributeError'> . Setting to 160x120 and resizing Requested resolution not supported: <class 'AttributeError'> . Setting to 160x120 and resizing Requested resolution not supported: <class 'AttributeError'> . Setting to 160x120 and resizing Requested resolution not supported: <class 'AttributeError'> . Setting to 160x120 and resizing Requested resolution not supported: <class 'AttributeError'> . Setting to 160x120 and resizing Requested resolution not supported: <class 'AttributeError'> . Setting to 160x120 and resizing Requested resolution not supported: <class 'AttributeError'> . Setting to 160x120 and resizing Requested resolution not supported: <class 'AttributeError'> . Setting to 160x120 and resizing 2017-10-11 16:26:37.844556: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations. 2017-10-11 16:26:37.844600: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations. 2017-10-11 16:26:38.117708: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:893] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero 2017-10-11 16:26:38.118064: I tensorflow/core/common_runtime/gpu/gpu_device.cc:955] Found device 0 with properties: name: GeForce GTX 1060 3GB major: 6 minor: 1 memoryClockRate (GHz) 1.759 pciBusID 0000:01:00.0 Total memory: 2.93GiB Free memory: 2.15GiB 2017-10-11 16:26:38.118356: I tensorflow/core/common_runtime/gpu/gpu_device.cc:976] DMA: 0 2017-10-11 16:26:38.118416: I tensorflow/core/common_runtime/gpu/gpu_device.cc:986] 0: Y 2017-10-11 16:26:38.118476: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060 3GB, pci bus id: 0000:01:00.0) /usr/local/lib/python3.5/dist-packages/tensorflow/python/ops/gradients_impl.py:95: UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory. "Converting sparse IndexedSlices to a dense Tensor of unknown shape. " Traceback (most recent call last): File "run_exp.py", line 131, in <module> main(sys.argv[1:]) File "run_exp.py", line 126, in main experiment.run(main_args[0]) IndexError: list index out of range

Is it because of the resolution not supported from vizdoom?

Hi, why the experiences contain nan?

https://github.com/IntelVCL/DirectFuturePrediction/blob/b4757769f167f1bd7fb1ece5fdc6d874409c68a9/DFP/future_predictor_agent_advantage.py#L86

Maybe this is just me, but I find this section as described in the paper highly confusing. What exactly is g? Is it the objective coefficients, the temporal coefficients (since it is supposed to have the same dimensionality as f, not as f_i), a combination of the two as this implementation assumes or the actual g * f?

It seems like there is a mistake in the target maker.

I guess we should substract the mean of the measurements in stead of the value of the measurement in itself.

targets[ns, :len(self.meas_to_predict), :] = ((meas[curr_future_steps][:, self.meas_to_predict] - meas_mean[:, self.meas_to_predict])/meas_std[:, self.meas_to_predict]).transpose() 

@dosovits If what I am saying is true, I would be happy to do a pull request.

Hi,
I installed all the dependent modules as required, including VizDoom, and import vizdoom works fine under test. To reproduce your experiments, the VizDoom path variable should be changed. However, I didn't figure out what does path stands for? Can anyone one give me an answer.
Thanks

When compiling on macOS, I get the following error :

Could not load the checkpoint checkpoints/2017_04_09_09_07_45

When I read the paper, they say that it works at discrete action space.

Is it also possible at continuous action space???

Really nice work, and thank you for releasing the code!

It was simple to find with a quick search, but including a link to the paper in the readme would be convenient.

Regards

Hi,

first of all, thanks for releasing the code!

Now regarding my issue, I cannot load any map in maps/ using set_doom_scenario_path. When I try it, it loads instead the first map of freedoom2.wad. If I try to open, let's say D1_basic.wad, with Slade, I get the error: Error opening D1_basic.wad: Unsupported or invalid Archive format.

Are the maps corrupted somehow? or do you avoid the issue in another way?

Thanks in advance!