bump torchao pin

.ci/scripts/test_llama_torchao_lowbit.sh

@@ -40,7 +40,6 @@ cmake --build cmake-out -j16 --target install --config Release
 
 # Install llama runner with torchao
 cmake -DPYTHON_EXECUTABLE=python \
-    -DCMAKE_PREFIX_PATH=$(python -c 'from distutils.sysconfig import get_python_lib; print(get_python_lib())') \
     -DCMAKE_BUILD_TYPE=Release \
     -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON \
     -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \

backends/vulkan/_passes/int4_weight_only_quantizer.py

@@ -7,7 +7,7 @@
 import torch
 import torch.nn.functional as F
 
-from torchao.quantization.GPTQ import _check_linear_int4_k
+from torchao.quantization.GPTQ.GPTQ import _check_linear_int4_k
 from torchao.quantization.unified import Quantizer
 from torchao.quantization.utils import groupwise_affine_quantize_tensor
 

examples/models/llama/CMakeLists.txt

@@ -116,6 +116,7 @@ endif()
 if(EXECUTORCH_BUILD_TORCHAO)
   # Currently only enable this on Arm-based Macs
   if(CMAKE_SYSTEM_NAME STREQUAL "Darwin" AND CMAKE_SYSTEM_PROCESSOR STREQUAL "arm64")
+    set(TORCHAO_BUILD_ATEN_OPS OFF)
     set(TORCHAO_BUILD_EXECUTORCH_OPS ON)
     set(TORCHAO_BUILD_CPU_AARCH64 ON)
     set(TORCHAO_ENABLE_ARM_NEON_DOT ON)

examples/models/llama/README.md

@@ -447,7 +447,6 @@ Next install the llama runner with torchao kernels enabled (similar to step 3.2
 
 ```
 cmake -DPYTHON_EXECUTABLE=python \
-    -DCMAKE_PREFIX_PATH=$(python -c 'from distutils.sysconfig import get_python_lib; print(get_python_lib())') \
     -DCMAKE_BUILD_TYPE=Release \
     -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON \
     -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \