Issues

• Don't use absolute file names in log.txt, metadata.json and probably other files (get rid of CWE-22)

  lightvector/KataGo#1255 · Filter by author KvanTTT opened on Sep 23, 2026

• Transformer model performance on Apple Neural Engine

  lightvector/KataGo#1251 · Filter by author MarcelGruenauer opened on Sep 5, 2026
  2

• Feature/API idea: temporarily prioritize a specific root move without clearing the current search tree

  lightvector/KataGo#1249 · Filter by author qiyi71w opened on Sep 2, 2026
  3

• ROCm backend on Windows fails to create MIOpen cache directory with non-ASCII username

  lightvector/KataGo#1248 · Filter by author Compreamateur opened on Sep 2, 2026
  5

• ROCm 7.2 + RDNA4: performance silently collapses (CK FMHA probe fails, ~2x slower)

  lightvector/KataGo#1245 · Filter by author nlzy opened on Aug 28, 2026
  5

• S

  lightvector/KataGo#1242 · Filter by author TUTU002791 opened on Aug 25, 2026

• [Discussion / Feedback] ONNX Backend: Benchmarks, Questions, and Issues

  lightvector/KataGo#1238 · Filter by author seniorfish opened on Aug 14, 2026
  16

• In katago selfplay, are the numSearchThreads and the value shown in the benchmark the same?

  lightvector/KataGo#1236 · Filter by author flowtensor835-byte opened on Aug 14, 2026
  2

• Could you give a rough estimate of the Elo difference between the new 15b weights and the 11b at the same visits? Which one is expected to have the advantage at the same compute time, and by how many Elo points? Thanks in advance!

  lightvector/KataGo#1233 · Filter by author flowtensor835-byte opened on Aug 10, 2026
  2

• When will the b15c1024 model in the Transformer series be updated

  lightvector/KataGo#1232 · Filter by author flowtensor835-byte opened on Aug 10, 2026

• User Test Report: b11c768 Transformer Model vs b28c512nbt on RTX 3080

  lightvector/KataGo#1224 · Filter by author GuangHui68 opened on Aug 3, 2026
  4

• [Feature Request] Transformer models are amazing on TRT, but OpenCL leaves 80%+ performance on the table – can we get a better general backend? (I can help test!)

  lightvector/KataGo#1220 · Filter by author Coderchen-001 opened on Jul 30, 2026
  7

• あ

  lightvector/KataGo#1217 · Filter by author kau17033 opened on Jul 26, 2026

• [News] Shin Jin-seo 2:1 KataGo with two handicaps, 10 years later AlphaGo

  lightvector/KataGo#1215 · Filter by author HackYardo opened on Jul 22, 2026
  2

• A Belated Correction for the 8x8 Opening Book

  lightvector/KataGo#1214 · Filter by author hhfhhfhfh1 opened on Jul 21, 2026
  4

• Is there a model that I can use it in ollama or other ?

  lightvector/KataGo#1213 · Filter by author ljzj2 opened on Jul 19, 2026
  2

• Add AMD HIP/Rocm support on Windows and Linux

  lightvector/KataGo#1212 · Filter by author KATAVM opened on Jul 14, 2026
  2

• Ошибка компиляции на Manjaro

  lightvector/KataGo#1210 · Filter by author Archyr opened on Jun 29, 2026
  3

• Incompatible with TensorRT 11

  lightvector/KataGo#1207 · Filter by author Galasnow opened on Jun 10, 2026
  5

• Data Distillation: Issues Encountered During Training

  lightvector/KataGo#1196 · Filter by author SakuraGoH opened on May 12, 2026

• Analysis engine's "field" property of error response is "rules" when it should be "komi"

  lightvector/KataGo#1193 · Filter by author kendfrey opened on May 9, 2026
  3

• runnnlayertests segmentation fault (zero pointer)

  lightvector/KataGo#1191 · Filter by author ThomasZoschke opened on May 5, 2026
  5

• Great Job

  lightvector/KataGo#1189 · Filter by author michito744 opened on Apr 20, 2026
  2

• Weird issue: PDA in range 7-8 causes no move to be played (only on FoxGTP, works fine on Lizzie)

  lightvector/KataGo#1186 · Filter by author SakuraGoH opened on Apr 17, 2026
  4

• Using KataGo’s Human SL model for non‑19x19 boards and human ranks (20k–9d)

  lightvector/KataGo#1184 · Filter by author chienbg2000 opened on Apr 16, 2026