-
I want to implement a MCTS planning algorithm, which makes frequent calls to `UPSequentialSimulator._get_applicable_actions()`. At each step of a MCTS rollout, the set of applicable actions must be ge…
-
#91 を参照 (やっつけ試作あり). 実装は清書中.
-
[gameplay-report.zip](https://github.com/magarena/magarena/files/2105568/gameplay-report.zip)
In turns 10 and 11, the AI decided to turn Blinkmoth Nexus into a creature for attacking and blocking r…
pshem updated
6 years ago
-
Allow simulations to run on different threads simultaneously and include a queue for neural net evaluation of leaf nodes in each thread.
-
Hey @zsunberg, I ran into *super* strange dependency problems with MCTS and POMDPPolicies because MCTS was in both the General registry and this registry. All of my MCTS issues were resolve by removin…
mossr updated
3 years ago
-
Sorry, I messed up and forgot the decks.
```
Traceback (most recent call last):
File "C:\Users\Ragowit\Documents\GitHub\mcts\hearthbreaker\uct.py", line 583, in UCTPlayGame
m = UCT(rootstate = …
-
I did some short test for self-play on the server. I needed to modify few things, so I created a branch 'self_play_speed_test'.
I used cProfiler for the profiling, here is the output of `python -m cP…
-
Trying to debug larger width environments (7 currently).
Things to try:
1. Different metric (Average Q-value from 2015 paper https://arxiv.org/pdf/1312.5602.pdf).
```
5.1 Training and Sta…
-
I understand the explanation of the MCTS AI "flailing" when it knows that it is losing (all courses being equally doom-filled). However, the game log below shows a situation where the AI (MCTS Level 8…
-
## Članovi tima
| Ime i prezime | Broj indeksa | Grupa |
| --------------- | -------------- | ------- |
| Nedeljko Vignjević | SW-59-2018 | 4 |
| Dalibor Malić | SW-50-2018 | 4 |
## Asiste…