issues
search
LeelaChessZero
/
lc0
The rewritten engine, originally for tensorflow. Now all other backends have been ported here.
GNU General Public License v3.0
2.37k
stars
523
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
fix rook positions after clearing board
#1942
borg323
closed
6 months ago
0
Make the sharpness limit in WDLRescale configurable, and fix the Elo --> Contempt calculation
#1941
Naphthalin
closed
6 months ago
2
Rename "win_percentage" UCI option to "expected_score"
#1940
TalvinJacobs
opened
7 months ago
1
Update readme to refer to latest 0.30 release
#1939
linrock
closed
7 months ago
0
simple multi-gpu configuration
#1938
borg323
opened
7 months ago
1
auto setting for max prefetch
#1937
borg323
opened
7 months ago
0
How to build Lc0 without OpenCL support (currently automatically picks it up)
#1935
BuddhiLW
closed
8 months ago
0
Create CITATION.cff
#1934
AlexisOlson
closed
8 months ago
0
Fix TaskWorkers range
#1933
skiminki
closed
8 months ago
0
Skip proactively wake up TaskWorkers for cpu backend
#1931
pdeng6
closed
8 months ago
14
auto setting of threads and task workers
#1930
borg323
closed
8 months ago
0
Improving the MUtility function
#1929
ContradNamiseb
opened
9 months ago
1
onnx2leela options to fix tf exported onnx models
#1928
borg323
closed
9 months ago
0
new network like this one don't work T2-768x15x24h-swa-5230000
#1927
Insanity1974
closed
10 months ago
0
Metal backend doesn't support iOS
#1926
nguyenpham
opened
10 months ago
3
sycl backend support for AMD, NVIDIA, and Intel
#1925
KateBlueSky
closed
9 months ago
23
Make leela2onnx Conv-nodes compatible with onnx2pytorch
#1924
patrik-ha
closed
9 months ago
6
No citation file
#1923
dpaleka
closed
8 months ago
3
fix numerical issues in onnx-dml fp16
#1921
borg323
closed
5 months ago
2
unknown error (../../src/neural/cuda/network_cudnn.cc:176)
#1920
kusayuzayushko
closed
5 months ago
2
onnx fix for opset 7
#1919
borg323
closed
11 months ago
0
fix Float16_t related onnx build issues
#1918
borg323
closed
11 months ago
0
Multiple head architecture for cuda backend
#1917
almaudoh
closed
5 months ago
0
Multihead arch metal
#1916
almaudoh
closed
5 months ago
0
Common changes for new multiple heads architecture.
#1915
almaudoh
closed
5 months ago
2
Fix layernorm epsilon for smolgen weights.
#1914
almaudoh
closed
11 months ago
0
added RMSNorm kernel
#1913
OfekShochat
opened
11 months ago
0
fix circleci macos build
#1912
borg323
closed
12 months ago
0
circleci tests
#1911
borg323
opened
12 months ago
0
Adding Improvements from EfficientZero
#1910
shermansiu
closed
12 months ago
3
Add hint to trust R3 certificate on Window
#1909
Ergodice
closed
12 months ago
0
fix build error with newer cuda versions
#1908
borg323
closed
1 year ago
0
rework selection of cuda build architecture #1905
#1907
whelanh
closed
1 year ago
1
rework selection of cuda build architecture
#1905
borg323
closed
1 year ago
3
Guidance on ensuring tensorflow_cc libraries are available when building on macOS
#1903
rschwiebert
closed
1 year ago
1
Access policy probabililties in engine
#1901
SimonBenhamou
closed
1 year ago
0
WDL in python binding
#1900
SimonBenhamou
closed
1 year ago
2
change default values for v0.30 release
#1899
Naphthalin
closed
1 year ago
2
handle play contempt mode with infinite search
#1898
borg323
closed
1 year ago
1
add support for onnxruntime-rocm to the onnx backend
#1897
borg323
closed
1 year ago
0
rename contempt perspective to contempt mode
#1895
borg323
closed
1 year ago
0
Drawscore simplification for Armageddon
#1894
Naphthalin
closed
1 year ago
0
fix empty contempt handling
#1893
borg323
closed
1 year ago
0
fix build.sh exit code
#1892
borg323
closed
1 year ago
0
Modify legacy TM to allow a bonus for first move if we expect shallow openings
#1891
Naphthalin
closed
1 year ago
0
Use back-off spin to wait for available searcher
#1890
pdeng6
closed
1 year ago
12
安卓lc0无法像围棋那样运行在npu上吗?在围棋中8gen2实力已经接近3060的水平,说不确定下一代8gen3就有3060的水平了。但只是他的精度有int8(8gen2的npu只支持int8和int4),但这是对围棋来说已经足够了。
#1889
ugvvff
closed
1 year ago
0
安卓lc0无法像围棋那样运行在npu上吗?在围棋中8gen2实力已经接近3060的水平,说不确定下一代8gen3就有3060的水平了。但只是他的精度有int8(8gen2的npu只支持int8和int4),但这是对围棋来说已经足够了。
#1888
ugvvff
closed
1 year ago
0
fix incllude files
#1886
borg323
closed
1 year ago
0
simplify cuda architectures used
#1885
borg323
closed
1 year ago
0
Previous
Next