COMMITS
/ examples/models/llama/source_transformation/quantize.py May 15, 2025
J
Refactor quantize.py functions to remove args
Jack committed
April 18, 2025
C
Fix bugs in executorch package
Chia-Jung Chang committed
April 17, 2025
April 7, 2025
S
[ET-VK][ez] Allow logit linear layer to be lowered to Vulkan (#9951)
Sicheng Stephen Jia committed
March 25, 2025
J
Switch to new ao quant api for 8da4w (#8501)
Jack committed
March 24, 2025
S
Updates torchao pin to enable shared embedding quantization (#9548)
Scott Roy committed
March 22, 2025
J
Fix xnnpack quantization discrepancy for non-fp32
Jack committed
March 20, 2025
M
Do not require checkpoint in quantize() unless it's gptq
Mengwei Liu committed
S
Bump torchao + add unit tests for torchao kernels (#9396)
Scott Roy committed
February 26, 2025
N
move import of VK 4-bit source quantizer into function
Nathanael See committed
S
Revert #8501 and #8624 (#8716)
Scott Wolchok committed
February 25, 2025
J
Switch to new ao quant api for 8da4w (#8501)
Jack committed
January 28, 2025
S
Assert quant_min/quant_max in embedding4bit (#7410)
Scott Roy committed
December 13, 2024
M
Add torchao mps lowbit ops to llama runner (#7037)
Manuel Candales committed
November 14, 2024
P
[ET-VK][Llama] Apply XNNPACK partitoner as well when lowering to Vulkan (#6857)
pytorchbot committed
November 11, 2024
S
Add torchao kernels to llama runner
Scott Roy committed
October 25, 2024
L
Replace torch.empty with torch.zeros
Lunwen He committed
October 16, 2024
M
Codemod examples/models/llama2 to examples/models/llama (#6302)
Mergen Nachin committed