PrivateGPT on windows 11 with 2 RTX3080 GPUs: only LLM chat works not QueryDocs

[maragem]

Hello,

i have successfully installed privateGPT on Windows 11 with 2 RTX 3080 GPUs. Driver Version: 546.33 CUDA Version: 12.3
the code runs well when using only the LLM chat. When i try to upload a document and use the QueryDocs it produced the following error:

RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:1 and cuda:0! (when checking argument for argument index in method wrapper_CUDA__index_select)

can anyone help please?
thanks

[Luis-Sarabando]

There is no support for multiGPU as far as i know