tlesort / learning-state-representation Goto Github PK

Probably some local path error?

Learning-State-Representation$ th script.lua
Test actuel : ./Log/19-10/Everything/
true
true
true
true
/home/natalia/torch/install/bin/luajit: /home/natalia/torch/install/share/lua/5.1/paths/init.lua:26: bad argument #1 to 'dir' (string expected, got nil)
stack traceback:
[C]: in function 'dir'
/home/natalia/torch/install/share/lua/5.1/paths/init.lua:26: in function 'files'
./Get_Images_Set.lua:9: in function 'images_Paths'
script.lua:91: in function 'train_Epoch'
script.lua:273: in main chunk
[C]: in function 'dofile'
...alia/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk
[C]: at 0x00406670

Also, regarding function naming:

1. What is supposed to do have-TODO() method?
2. Should the method doStuff_* [temp, caus, etc.) be called something more intuitive?, e.g. perhaps backpropagate() or getLossAndGradient()?

As GPU mode is inaccessible for me at the moment, in CPU only mode I get the following running script.lua in the useCUDA flag CPU version of this file in :
https://github.com/Mathieu-Seurin/baxter_representation_learning_3D/script.lua

Computing performance...
Optimization with Priors backpropagation: in CPU mode this takes forever..
getRandomBatch for sample...
Applying optimization with priors backprop...
/home/natalia/torch/install/bin/luajit: /home/natalia/torch/install/share/lua/5.1/nn/THNN.lua:110: $ Torch: not enough memory: you tried to allocate 0GB. Buy new RAM! at /home/natalia/torch/pkg/torch/lib/TH/THGeneral.c:270
stack traceback:
[C]: in function 'v'
/home/natalia/torch/install/share/lua/5.1/nn/THNN.lua:110: in function 'SpatialConvolutionMM_updateOutput'
...ia/torch/install/share/lua/5.1/nn/SpatialConvolution.lua:79: in function 'func'
.../natalia/torch/install/share/lua/5.1/nngraph/gmodule.lua:345: in function 'neteval'
.../natalia/torch/install/share/lua/5.1/nngraph/gmodule.lua:380: in function 'forward'
./priors.lua:17: in function 'doStuff_temp'
...lia/dream/baxter_representation_learning_3D/printing.lua:45: in function 'Print_performance'
script.lua:110: in function 'train_Epoch'
script.lua:307: in main chunk
[C]: in function 'dofile'
...alia/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk
[C]: at 0x00406670

Any obvious parts that can be made more sequential to handle only required images at each batch to optimize for cpu mode only?

Seems like the model
https://github.com/TLESORT/Learning-State-Representation/blob/master/models/topTripleFM_Split.lua
is only made for 1D representation learning?

The error produced:

`
baxter_representation_learning_3D$ th train.lua
Running main script with useCUDA flag: false
Running main script with useSecondGPU flag: false
nb_parts per batch: 50 LearningRate: 0.001 BatchSize: 1. Using data folder: ./baxter_data
Get_HeadCamera_View_Files(Path: ./baxter_data
list_folders_images=
{
1 : "/home/natalia/dream/baxter_representation_learning_3D/baxter_data/record_0/recorded_cameras_head_camera_2_image_compressed"
2 : "/home/natalia/dream/baxter_representation_learning_3D/baxter_data/record_1/recorded_cameras_head_camera_2_image_compressed"
3 : "/home/natalia/dream/baxter_representation_learning_3D/baxter_data/record_10/recorded_cameras_head_camera_2_image_compressed"
4 : "/home/natalia/dream/baxter_representation_learning_3D/baxter_data/record_2/recorded_cameras_head_camera_2_image_compressed"
5 : "/home/natalia/dream/baxter_representation_learning_3D/baxter_data/record_3/recorded_cameras_head_camera_2_image_compressed"
6 : "/home/natalia/dream/baxter_representation_learning_3D/baxter_data/record_4/recorded_cameras_head_camera_2_image_compressed"
7 : "/home/natalia/dream/baxter_representation_learning_3D/baxter_data/record_5/recorded_cameras_head_camera_2_image_compressed"
8 : "/home/natalia/dream/baxter_representation_learning_3D/baxter_data/record_6/recorded_cameras_head_camera_2_image_compressed"
9 : "/home/natalia/dream/baxter_representation_learning_3D/baxter_data/record_7/recorded_cameras_head_camera_2_image_compressed"
10 : "/home/natalia/dream/baxter_representation_learning_3D/baxter_data/record_9/recorded_cameras_head_camera_2_image_compressed"
}
Running test: 1 . Saving in Log_Folder: ./Log/10-19/Everything/...
Which priors are being applied?:
true
true
true
true
images Path variable suddenly becomes nil here if not reset!: ( WHY?!)
./baxter_data
Loaded images from Path: ./baxter_data
txt_reward_test (button pressed)=/home/natalia/dream/baxter_representation_learning_3D/baxter_data/record_9/recorded_button1_is_pressed.txt
txt_test (state) =/home/natalia/dream/baxter_representation_learning_3D/baxter_data/record_9/recorded_robot_limb_left_endpoint_state.txt
list_truth ={}
show_figure for truth : { <userdata 1>, <userdata 2>, <userdata 3>, <userdata 4>, <userdata 5>, <userdata 6>, <userdata 7>, <userdata 8>, <userdata 9>, <userdata 10>, <userdata 11>, <userdata 12>, <userdata 13>, <userdata 14>, <userdata 15>, <userdata 16>, <userdata 17>, <userdata 18>, <userdata 19>, <userdata 20> } and Data_test:{
Infos = {
dx = { 0.627270879639 },
dy = { 0.433297297588 },
dz = { 0.31086588434 },
reward = { 0 }
},
images = {}
} Log_Folder./Log/10-19/Everything/ Data_test.Infos: 0
show_figure for list of size: 20 and Infos.reward of size: 1 Name: ./Log/10-19/Everything/The_Truth.Log Variable_Name: Truth
/home/natalia/torch/install/bin/luajit: ...lia/dream/baxter_representation_learning_3D/printing.lua:206: error list are not same lenght
stack traceback:
[C]: in function 'assert'
...lia/dream/baxter_representation_learning_3D/printing.lua:206: in function 'show_figure'
train.lua:135: in function 'train_Epoch'
train.lua:299: in main chunk
[C]: in function 'dofile'
...alia/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk
[C]: at 0x00406670

`

which is equivalent to the following, using the print() method:
show_figure for truth : { 1 : DoubleTensor - size: 3 2 : DoubleTensor - size: 3 3 : DoubleTensor - size: 3 4 : DoubleTensor - size: 3 5 : DoubleTensor - size: 3 6 : DoubleTensor - size: 3 7 : DoubleTensor - size: 3 8 : DoubleTensor - size: 3 9 : DoubleTensor - size: 3 10 : DoubleTensor - size: 3 11 : DoubleTensor - size: 3 12 : DoubleTensor - size: 3 13 : DoubleTensor - size: 3 14 : DoubleTensor - size: 3 15 : DoubleTensor - size: 3 16 : DoubleTensor - size: 3 17 : DoubleTensor - size: 3 18 : DoubleTensor - size: 3 19 : DoubleTensor - size: 3 20 : DoubleTensor - size: 3 } and Data_test (.Infos): { Infos : { dy : { 1 : 0.433297297588 } reward : { 1 : 0 } dx : { 1 : 0.627270879639 } dz : { 1 : 0.31086588434 } } images : {...} }
Where <userdata 1>, <userdata 2> signify the internal representation of Torch datatypes.