COMMITS
/ kernels/quantized/cpu/op_dequantize.cpp January 28, 2025
S
Execute on exec_aten namespace deprecation from #5296 (#7950)
Scott Wolchok committed
December 17, 2024
R
Disable c10::optional macros in executorch
Richard Barnes committed
December 2, 2024
P
[Executorch][quant] Optimize per channel dequantize
pytorchbot committed
November 15, 2024
J
Add UInt16 support to Q and DQ
Jacob Szwejbka committed
November 5, 2024
R
c10::optional -> std::optional
Richard Barnes committed
October 9, 2024
S
Support quantizing to and dequantizing from uint16_t (Bits16) (#5839)
Shen Xu committed
October 2, 2024
K
Add quantized kv cache to llama (#5664)
Kimish Patel committed
September 26, 2024
K
Fix dequantize per channel to handle double scale type (#5524)
Kimish Patel committed
September 12, 2024
D
Migrate RuntimeContext users to KernelRuntimeContext (#5270)
Dave Bort committed
June 18, 2024
T
Fix dequantize_per_channel for single dimension input tensor (#3867)
Tarun Karuturi committed
May 31, 2024
S
Manual LICM for numel() in quantize ops (#3785)
Scott Wolchok committed
S
Manual LICM for mutable/const_data_ptr calls in quantize ops (#3784)
Scott Wolchok committed
May 16, 2024
S
Fix -Wdeprecated-defintions + -Werror data_ptr complaints in kernels (#3639)
Scott Wolchok committed
March 12, 2024
M
optional zero points on dequantize per channel (#2364)
Manuel Candales committed
December 8, 2023
M
Add dequantize_per_channel.out kernel (#1375)
Mengwei Liu committed
September 25, 2023
V
Fix %hhd format specifiers as PRId8 (#470)
Vincent Lee committed
September 22, 2023
V
Fix build breaks from llvm-17 (#461)
Vincent Lee committed
September 13, 2023
R
Support torch.int32 as a dtype for quantize and dequantize (#289)
Riley Dulin committed
July 31, 2023
F
Initial commit
facebook-github-bot committed