compile error"Vulkan backend does not support quantization at the moment"

🐛 Describe the bug

when I compile the pte file from llama-7b-chat as indicated by "https://pytorch.org/executorch/stable/build-run-vulkan.html", I find that the generated ptr file size is too big to run on the edge devices. So I tried to use quantization to the Vulkan bankend. the command is as follows:
python -m examples.models.llama2.export_llama --disable_dynamic_shape --vulkan -kv --use_sdpa_with_kv_cache --checkpoint ~/Llama-2-7b-chat/consolidated.00.pth --params ~/Llama-2-7b-chat/params.json -d fp32 -X -qmode 8da4w --group_size 128 --max_seq_length 1024
then it was generated the error as follows:
Vulkan backend does not support quantization at the moment
I tried to generate the pte file at executor v0.4.0
Conld you help me fix the issue?
Thank you in advanced

Versions

executorvh v0.4.0

cc @SS-JIA @manuelcandales

Start with the examples.models.llama2.export_llama entry point and the Vulkan build instructions linked in the report. Reproduce the command with executor v0.4.0 and trace where the explicit quantization rejection occurs. Done means the quantized Llama export completes for Vulkan and produces a PTE suitable for the reported edge-device use case.