Support for QWEN and Baichuan2 models

[sorasoras]

Feature request

recently, https://github.com/ggerganov/llama.cpp has add support for both QWEN and Baichuan2.
It has added QWEN at 1610.
ggerganov/llama.cpp#4281
I have look up the Nomic Vulkan Fork of LLaMa.cpp,
it does have support for Baichuan2 but not QWEN, but GPT4ALL itself does not support Baichuan2.

Motivation

I failed to load baichuan2 and QWEN models, GPT4ALL supposed to be easy to use

Your contribution

Not quite as i am not a programmer but i would look up if that helps

[cebtenzzre]

Baichuan2 should be supported in the next release - the current release isn't using the latest version of our llama.cpp fork.

Qwen got merged upstream a little too late, but it should be supported here soon.

[AdkinsHan]

There is a problem. When I use Qwen, no matter the graphics card or CPU specified by my device, it has been calling the CPU and memory for calculation instead of calling the GPU.
mode_name="qwen1_5-14b-chat-q8_0.gguf"
gpt4all version 2.2.1.post1
https://github.com/QwenLM/Qwen


When I use it, I see that the memory and CPU are being called, but the video memory usage has not changed.

[AdkinsHan]

When I use other GGUF models I can see the following output
llama.cpp: using Vulkan on NVIDIA GeForce RTX 3070

[cebtenzzre]

As of #2005, Qwen, Qwen2, and Baichuan all have GPU support enabled on both the Metal and Vulkan backends.

[cebtenzzre]

This is implemented in the v2.7.2 release.