issues
search
huangeddie
/
MuZeroGoJax
Mu Zero Go implemented with JAX and GoJAX
MIT License
9
stars
0
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Do past game moves help with attention?
#296
huangeddie
opened
10 months ago
0
Pit model against an external computer AI
#295
huangeddie
opened
10 months ago
0
Aigagror/issue291
#294
huangeddie
closed
1 year ago
0
Save flags in model dir
#293
huangeddie
closed
1 year ago
0
Augment game data instead of trajectories
#292
huangeddie
opened
1 year ago
0
Use PyDrive for more persistent storage
#291
huangeddie
closed
1 year ago
2
Aigagror/issue288
#290
huangeddie
closed
1 year ago
0
Make flag for first training step optional to save time
#289
huangeddie
opened
1 year ago
0
Trajectory batch buffer
#288
huangeddie
closed
1 year ago
0
Fix broken tests
#287
huangeddie
closed
1 year ago
0
Simple custom sequential search tree exploration
#286
huangeddie
opened
1 year ago
1
learning rate schedule
#285
huangeddie
opened
1 year ago
0
remove all legacy instances of decode models
#284
huangeddie
closed
1 year ago
0
Measure qval entropy instead of qcomplete
#283
huangeddie
closed
1 year ago
0
Measure qcomplete entropy
#282
huangeddie
closed
1 year ago
0
Fix eval and plot traj dtype bug
#281
huangeddie
closed
1 year ago
0
Fix log_softmax to float32
#280
huangeddie
closed
1 year ago
0
Use Jax mixed precision library
#279
huangeddie
closed
1 year ago
0
Remove legacy value models
#278
huangeddie
closed
1 year ago
0
Rename value to final area
#277
huangeddie
opened
1 year ago
0
Plot predicted final areas
#276
huangeddie
closed
1 year ago
0
Save and load dataframe
#275
huangeddie
closed
1 year ago
0
Aigagror/issue270
#274
huangeddie
closed
1 year ago
0
Log metrics before first train step
#273
huangeddie
closed
1 year ago
0
Improve get_benchmarks
#272
huangeddie
closed
1 year ago
0
Test bfloat16 again
#271
huangeddie
closed
1 year ago
0
Insert end state area decoder
#270
huangeddie
closed
1 year ago
0
Initialize model policy to centered normal distribution
#269
huangeddie
closed
1 year ago
0
test train module
#268
huangeddie
opened
1 year ago
0
Copy gojax source code in
#267
huangeddie
closed
1 year ago
0
Split GitHub Actions into smaller test components
#266
huangeddie
closed
1 year ago
0
Refine postsubmit
#265
huangeddie
closed
1 year ago
0
Split train module into train_manager and train_step
#264
huangeddie
closed
1 year ago
0
Replace decode loss with area loss
#263
huangeddie
closed
1 year ago
0
Buffer donation on multi_train_step_fn
#262
huangeddie
closed
1 year ago
0
Save model every so often
#261
huangeddie
closed
1 year ago
0
Move trained models to Git Large Files Storage
#260
huangeddie
closed
1 year ago
0
Speed up tests by removing or making them more efficient
#259
huangeddie
closed
1 year ago
0
Plot elo over training time
#258
huangeddie
closed
1 year ago
0
Assert distributed params and opt state are equal at the end of training
#257
huangeddie
closed
1 year ago
0
pmap elo eval
#256
huangeddie
closed
1 year ago
0
Directory for flag config files
#255
huangeddie
closed
1 year ago
0
train step traces three times?!
#254
huangeddie
closed
1 year ago
0
pmap self_play, and update_step
#253
huangeddie
closed
1 year ago
0
Benchmark base models on image classification
#252
huangeddie
opened
1 year ago
0
Policy loss on hypothetical embeddings
#251
huangeddie
opened
1 year ago
0
Make max max action capacity dynamically equal to max actions sampled
#250
huangeddie
closed
1 year ago
0
Fused MB Conv
#249
huangeddie
closed
1 year ago
0
Dropout
#248
huangeddie
closed
1 year ago
0
Plot end states
#247
huangeddie
closed
1 year ago
0
Next