GGML Assert error

[ADXXI1590]

GGML_ASSERT: C:\Users\circleci.PACKER-64370BA5\project\gpt4all-backend\llama.cpp\ggml.c:4411: ctx->mem_buffer != NULL
not getting any prompt to enter the query?
instead getting the above assertion error? can anyone help with this?

[Om-GG]

I have the same issue.

[jhoughtelin]

Increase the amount of memory available. I had the same issue trying to run the python binding of gpt4all. It worked after I provisioned a system with more than 16G of memory.
Slightly different but #90 is also suggesting that the system does not have enough memory.

[Om-GG]

Yep, I gave it 8 gb and it started moving, good to know that 16 may be a good number to aim for. It doesn't seem that 8gb would be anywhere near enough, as it moves quite slowly and takes a long time to reply to queries.

[rwcitek]

I had the same issue.

I was using a 2 GB cloud instance to build a Docker image with privateGPT. The build worked. Here's a link to the Dockerfile:

https://github.com/rwcitek/privateGPT/blob/docker/docker/Dockerfile

However, when I tried to run privateGPT.py in the instance, it would core dump. Changing the cloud instance to 16GB enabled the container to run. This is the Docker run command that I used:

docker run --rm -it rwcitek/privategpt:2023-06-04 python3 privateGPT.py

Since I did my development using GitHub CodeSpaces, I modified my devcontainer.json file to specify 16GB.

https://github.com/rwcitek/privateGPT/blob/docker/.devcontainer/devcontainer.json

I have not tried it on anything between 2 GB and 16 GB.

[selcukbeyhan]

same error. I saw multiple reports of this issue:
#90
#186

Is this something part of the model? or something related to this "circleci" user "C:\Users\circleci.PACKER-64370BA5"?