infiniflow
[Bug]: The progress cannot advance in MacOS (showing ⚠️ARM64)
Is there an existing issue for the same bug?
- [X] I have checked the existing issues.
Branch name
main
Commit ID
88072b1e900cf69716038bba77087e19f38ee45e
Other environment information
No response
Actual behavior
When installing ragflow, the ARM64 architecture has already been configured, but after running Docker, it still shows ⚠️ amd64, and the CPU usage reaches nearly 200%. The performance is reported to be extremely low. Additionally, while parsing DOC files, the progress cannot advance. It seems that this open-source project cannot function effectively on macOS.
Expected behavior
No response
Steps to reproduce
I just follow the standard steps for MacOS. And I also use a updated requirements_arm.txt and Dockerfile.arm in a closed case"[Macosx M1 not supported?](https://github.com/infiniflow/ragflow/issues/1164#top) #1164".
Similar situations with the same bug(⚠️AMD64)
Additional information
No response
You can try dev-slim version of docker image since it do not handle embeding inference itself.
I have
You can try dev-slim version of docker image since it do not handle embedding inference itself.
I have tried all versions, same situations ⚠️AMD64 and embedding doesn't work.
For slim version of RAGFlow, it dose not handle embedding procedure itself. You need to use Ollama to deploy an embedding inference.
M3 chip, installation fails, error message rosetta error: Rosetta is only intended to run on Apple Silicon with a macOS host using Virtualization.framework with Rosetta mode enabled % , is there a similar situation?
it work on M2 chip , I build docker image from source code. follow these steps :
conda env
git clone https://github.com/infiniflow/ragflow.git
cd ragflow/
pip3 install huggingface-hub nltk
python3 download_deps.py
docker build --build-arg ARCH=arm64 -f Dockerfile.slim -t infiniflow/ragflow:dev-slim .
docker compose -f docker-compose.yml -p ragflow up -d
We have difficulty on building multi-arch image with CI. So we will not publish arm64 image officially in the near future.
However you can build it by yourself on a linux/arm64 or darwin/arm64 host.
