SavchenkoValeriy/wazuhl

Issues

• Implement "peeping" policy

  #39 opened 5 years ago by SavchenkoValeriy
  0

• Implement Double Q learning algorithm

  #38 opened 5 years ago by SavchenkoValeriy
  0

• Implement normalized timer

  #15 opened 7 years ago by SavchenkoValeriy
  0

• Add --train-wazuhl option to clang

  #26 opened 7 years ago by SavchenkoValeriy
  0

• Get rid of caffe debug output

  #37 opened 7 years ago by SavchenkoValeriy
  0

• Implement basic neural network for Q estimation

  #28 opened 7 years ago by SavchenkoValeriy
  0

• Implement infrastructure to make running passes according to polices possible

  #34 opened 7 years ago by SavchenkoValeriy
  0

• Link with Caffe without setting LD_LIBRARY_PATH

  #33 opened 7 years ago by SavchenkoValeriy
  1

• Implement experience replay

  #36 opened 7 years ago by SavchenkoValeriy
  0

• Add unittests for Q-function wrapper

  #35 opened 7 years ago by SavchenkoValeriy
  0

• Find out the way to store caffe model alongside with installed llvm

  #29 opened 7 years ago by SavchenkoValeriy
  0

• Implement Wazuhl's environment

  #32 opened 7 years ago by SavchenkoValeriy
  0

• Try LSTM on Instruction sequence as Q-function estimator

  #31 opened 7 years ago by SavchenkoValeriy
  0

• Add N-grams and TF-IDF based on Instructions as features

  #30 opened 7 years ago by SavchenkoValeriy
  0

• Figure out the way how to compile and run one test from llvm-test-suite

  #25 opened 7 years ago by SavchenkoValeriy
  1

• Pick a library to make a Q-function estimator NN

  #8 opened 7 years ago by SavchenkoValeriy
  1

• Add proper header blocks to .h files

  #27 opened 7 years ago by SavchenkoValeriy
  0

• Find out the best way of using graphs (call graph and CFG) as features

  #9 opened 7 years ago by SavchenkoValeriy
  1

• Pick tests to train wazuhl

  #1 opened 7 years ago by SavchenkoValeriy
  3

• Implement a pass collecting interesting info about IR

  #6 opened 7 years ago by SavchenkoValeriy
  0

• Make OW do nothing when configured without LLVM_ENABLE_WAZUHL

  #24 opened 7 years ago by SavchenkoValeriy
  0

• Implement action dependency analisys

  #13 opened 7 years ago by SavchenkoValeriy
  1

• Filter out statistics that will never be non-zero

  #23 opened 7 years ago by SavchenkoValeriy
  0

• Make -OW run without crashes on any random sequence of actions taken

  #22 opened 7 years ago by SavchenkoValeriy
  0

• Implement collection of statistics of the pass performed

  #19 opened 7 years ago by SavchenkoValeriy
  0

• Move the creation of a list of all possible actions to compile-time

  #21 opened 7 years ago by SavchenkoValeriy
  0

• Implement a penalty-giving for picking a pass with unsatisfied requirements

  #20 opened 7 years ago by SavchenkoValeriy
  0

• Implement Q-learing algorithm

  #14 opened 7 years ago by SavchenkoValeriy
  1

• Implement new policy

  #18 opened 7 years ago by SavchenkoValeriy
  0

• Research the possibility of using multi-step DQN

  #17 opened 7 years ago by SavchenkoValeriy
  0

• Implement timer fine

  #16 opened 7 years ago by SavchenkoValeriy
  0

• Pick a library to perform reinforcement learning

  #3 opened 7 years ago by SavchenkoValeriy
  1

• Get rid of printing passes from possible actions list

  #12 opened 7 years ago by SavchenkoValeriy
  1

• Add Function, Loop, etc. passes to all possible actions

  #10 opened 7 years ago by SavchenkoValeriy
  0

• Fix OW performance

  #11 opened 7 years ago by SavchenkoValeriy
  0

• Introduce -OW option in clang

  #4 opened 7 years ago by SavchenkoValeriy
  1

• Implement decision making foundation

  #7 opened 7 years ago by SavchenkoValeriy
  0

• Implement training infrastructure

  #5 opened 7 years ago by SavchenkoValeriy
  0

• Think of an architecture to make training process possible

  #2 opened 7 years ago by SavchenkoValeriy
  0