[BUG] Error running LLaMA2_7B_Chat_Quantized on 8gen3 device.

[LLIKKE]

Hi, on LLaMA2_7B_Chat_Quantized, I noticed that the compile job on the ai hub is the QNN : v2.27.0.240926142112_100894 version, but no matter if I use 2.27.7 or 2.27.0 I get the same error!

HNBVL-AN00:/data/local/tmp $ cd llama2_7b_qnn/
user<|end_header_id|>\n\nWhat is France's capital?<|eot_id|><|start_header_id|>assistant<|end_header_id|>"            <
Using libGenie.so version 1.1.0

[WARN]  "Unable to initialize logging in backend extensions."
[INFO]  "Using create From Binary"
[INFO]  "Allocated total size = 300255744 across 8 buffers"
[ERROR] "Could not create context from binary for context index = 1 : err 4000"
[ERROR] "Create From Binary FAILED!"
[ERROR] "Failed to free device: 14003"
[ERROR] "Device Free failure"
Failure to initialize model
Failed to create the dialog.

I follow https://github.com/quic/ai-hub-apps/tree/main/tutorials/llm_on_genie. using 8gen3. But llama3_2 3b model runs well.