- Ver. 1.0.0
- Ver. 2.0.0
- Half resolved the bug n°4 (Check the
Bug List.md files
to see the explication of te fix)
- Implementation of the test of the Q-Table after training. For a 100 episodes you can try your Q-Table to see if the update is good or not.
- If the results is upper or equal than 50% it is a good update between 33% and 49% it is not good at it could be, between 25% and 33% that's not a good update as well and less than 25% that's not a good update.
- Adding new datas obtention:
longest_sequence
: List of states in the longer episode that doesn't reach the goallongest_best_sequence
: List of states in the longest episode that reach the goalshortest_sequence
: List of states in the shortest episode that doesn't reach the goalrecurent_sequence
: Number of the episodes that the agent done the same sequence to reach the goal with the best sequence
- Adding a message for a success rate:
- Of 100%
- Between 80% and 99%
- Between 50% and 79%
- Bug/Problem found and listed (For more informations about the Bugs/Problems checks the
Bug list.md
in thedoc
folder):- New bugs/problems:
- n°5, n°6
- Problem not fixed yet:
- n°1
- n°2
- n°3
- n°5
- n°6
- Problem fixed:
- n°4 (fixed in the patch 1.0.1)
- New bugs/problems:
- Bug/Problem fixed (not listed):
- Correction of the launch of the test of the updated Q-Table
- Calculation of espsilon decay (calcul detail:
1/episodes
) - Input value of epsilon in the console
- Bug fix(not listed)
- Q-Table training sucess rate calculation
- Upadate Gym package pass to 1.0.0 alpha version
- Bug/Problem fix (listed) (For more informations about the Bugs/Problems checks the
Bug list.md
in thedoc
folder):- n°1
- n°2
- n°3
- n°5
- n°7
- Adding an example of Training + test results (Check the
4x4 Results.txt
file in the4x4
folder in theQ-Learning-Results
folder) - Bug/Problem fix (listed) (For more informations about the Bugs/Problems checks the
Bug list.md
file in thedoc
folder):- n°6
- n°8
- Re-layout of the
README.md
,Bug List.md
,Patch Note.md
files - Replacing by an another example the example of final result for the 4x4 map in the
4x4 Results.txt
+ adding a picture of the outcomes of the training + test in the4x4
folder in theQ-Learning-Results
folder - Bug/Problem fix (listed) (For more informations about the Bugs/Problems checks the
Bug list.md
file in thedoc
folder):- n°9
- Bug/Problem fix (not listed):
total_actions
are all actions of the training + test part. To change this we reset the value at the start of the test part.
- BRAND NEW FUNCTIONALITY: Q-Injection
The Q-Injection is for the training and/or the test of Q-Table Like:
- Randomized Q-Table
- Trained Q-Table
- Little updated Q-Table
These Q-Tables are in the QInjection.py
file
(For more information about check the injection.md
file in the Tools
folder)
- Bug/Problem found and listed (For more informations about the Bugs/Problems checks the
Bug list.md
in thedoc
folder):- n°10
- Adding the Q-Injection for the 2x2 (Check the
Q-injection.md
in theTools
folder to find more informations about the Q-Injection) - Re-layout of
maps.txt
nowmaps.md
to allow a better copy -> paste of the maps (Check themaps.md
in theTools
folder to see the re-layout)