ENOENT: No such file or directory, open 'binariesGithubRelease.json'

[calycekr]

Bug description

After running npm run build and then node build, I encounter an internal server error (500) due to a missing file. The error message indicates that the file /app/build/server/llama/binariesGithubRelease.json cannot be found.

Context

Logs

{"level":50,"time":1744627276526,"pid":871547,"hostname":"devserver","locals":{"sessionId":"02bc8501179756c0f2526e17d0f426af539c41df1a8b1eeb7df2e5793ce8eefe"},"url":"https://***/","params":{},"request":{},"message":"Internal Error","error":{"errno":-2,"code":"ENOENT","syscall":"open","path":"/app/build/server/llama/binariesGithubRelease.json"},"errorId":"8cc23c3a-4ad2-4e04-8364-bdb80c5fda99","status":500,"stack":"Error: ENOENT: no such file or directory, open '/app/build/server/llama/binariesGithubRelease.json'"}

Notes

I am using a proxy environment.

This issue seems to be related to node-llama-cpp.
The file binariesGithubRelease.json is expected to be present in the specified path but is not found.

[menahem-borges-rodrigues]

Same bug here.

[nsarrazin]

Should be better now with the latest commit! I added a step that builds llama.cpp in npm run build and it has to run twice which is super annoying and slows things down but at least it should work again.

[calycekr]

@nsarrazin Whenever I run npm run build, the llama.cpp source build is also executed. Is this the intended behavior?
It would be helpful to have an option to not use the local model. Could you please consider this?

[nsarrazin]

Yep I'll add an option to skip it

[Faugermire]

@nsarrazin Hello, I seem to be encountering this error when attempting to host ChatUI through Docker (the image with MongoDB included: ghcr.io/huggingface/chat-ui-db:sha-495869a at https://github.com/orgs/huggingface/packages/container/chat-ui-db/413126380?tag=sha-495869a, and I am using the ARM64 variant). Is this a misconfiguration on my end, or does the image just need to be updated in the hub?