KataGo

KataGo copied to clipboard

bug report

4

hello lightvector ! i am trying to play with KataGo using play.py ,however i discovered: "gamestate.py", line 90, in get_input_features bin_input_data = np.zeros(shape=[1]+features.bin_input_shape, dtype=np.float32) ^^^^^^^^^^^^^^^^^^^^^^^^ AttributeError: 'Features' object has no...

yehu3d

CUDA 12.5 build produces visits/s = -nan(ind) even with compatible cuDNN and model

1

Hi, I'm having trouble getting the CUDA 12.5 version of KataGo to work on Windows 11. The binary runs, but all benchmark tests fail with visits/s = -nan(ind) output —...

Immow

[feature request] supporting multiple backends simultaneously

1

Hi. Currently, KataGo has common interface for backends, and the cache variable USE_BACKEND determines which implementation is used. On Apple Silicon, two (or three) backends, Neural Engine, GPU, (and CPU),...

y-ich

【Rule Proposal】Unilateral-Button: enable even komi (like 6-point komi) under Chinese/Area rules

7

**Button**: Whether a half-point is awarded to the first player to be able to pass. Then, how about to introduce "Unilateral-Button" rule? **Unilateral-Button**: A half-point komi is only awarded to...

poptangtwe

TensorRT issue on RTX5090

16

Hi, Downloaded the latest v1.15.3 katago. OpenCL works fine. Latest NVIDIA driver. But when I try to use TensorRT, it shows as below issue: Running quick initial benchmark at 16...

blademeng

Question about using updated b28 network file with older CoreML .mlpackage

1

I’m running the CoreML version of KataGo on a Mac with M2 chip. Since the latest CoreML-supported version with 19x19 input appears to be from v1.15.1 (with the .bin.gz network...

Rick130130

Integrating attention and feature vectors?

2

I want to train my own net and make some additions onto the katago framework. Using my very limited AI knowledge, I theorized that a 2d pos encoded attention layer...

Centaurea-Platinum

How to use performance of the "FP16 Tensor TFLOPs with FP16 Accumulate"

2

I noticed that there are so many graphics cards with Tensor Cores, peak performance of the "FP16 Tensor TFLOPs with FP16 Accumulate" is twice than peak performance of the "FP16...

guitanj

Cuda graph implementation on trt backend

2

zjuwyz

Still unresolved and worsening problems

16

A problem that AI cannot solve at all, which arose in a professional real-life situation.   This is due to the inability to effectively solve loose ladders—a long-standing issue...

michito744