issues
search
lowrollr
/
turbozero
fast + parallel AlphaZero in JAX
Apache License 2.0
85
stars
7
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
fix: version bumps
#20
lowrollr
closed
2 days ago
0
Detecting CUDA/cuDNN
#19
r-wedeen
closed
2 days ago
3
Confusion around jit in train loop
#18
ConstantinRuhdorfer
closed
2 months ago
2
Hey, Could you please share your baseline runs?
#17
DuaneNielsen
closed
3 months ago
1
The key differences between this work and the implementation of alphazero in PGX
#16
CDM1619
closed
1 month ago
2
Allow custom data augmentation functions to generate additional training samples stored in replay memory
#15
lowrollr
closed
8 months ago
0
Add docstrings to all functions/classes, general code cleanup
#14
lowrollr
closed
8 months ago
0
Allow for rendering test games, user-defined custom baselines + other small Tester improvements
#13
lowrollr
closed
8 months ago
0
Add multi-gpu support for self-play, training, and testers
#12
lowrollr
closed
8 months ago
0
Batch MCTS is needed !!!
#11
Nightbringers
closed
2 months ago
7
allow for user-specified data augmentation in Trainer
#10
lowrollr
closed
8 months ago
1
allow for running w/ multiple gpus and provide an example
#9
lowrollr
closed
8 months ago
1
bug
#8
Nightbringers
closed
9 months ago
34
speed issue
#7
Nightbringers
closed
10 months ago
1
jax
#6
lowrollr
closed
10 months ago
1
AlphaZero+MCTS: Visit probabilities for invalid actions can be non-zero
#5
bubble-07
closed
2 months ago
3
fixed dirichlet spelled dirilecht
#3
DavideTr8
closed
1 year ago
0
Dirilecht instead of dirichlet in mcts
#2
DavideTr8
closed
1 year ago
1
LazyZero-based training sample commands fail with "invalid multinomial distribution"
#1
bubble-07
closed
1 year ago
7