Stuck Loading

[arjuna-dev]

Stuck Loading

The app gets stuck loading on any query. Never got past it.

Steps To Reproduce
Steps to reproduce the behavior:

1. Open the app
2. Select model (using alpaca-7b-native-enhanced from hugging face, file: ggml-model-q4_1.bin)
3. Make query

Expected behavior
I should get an answer after a few seconds (or minutes?)

Screenshots

OS

• OS: macOS Ventura 13.1 (intel)

[01miru]

If you are using the docker version, you must add the command before compiling alpaca in Dockerfile (as in the example):

RUN export LD_LIBRARY_PATH=/usr/local/lib64:$LD_LIBRARY_PATH

[arjuna-dev]

Hi, I am using the desktop Electron app for Mac intel

[232344235342]

Same issue for me on OSX ventura 13.2 intel

[rcooknet]

I have the same issue. Electron wrapper. Any query just shows a loading spinner, and I time out after 5-10 minutes.

macOS 12.5 Monterey
M1 Max 32GB ram

[nafdez]

Hi, same issue but on windows. Never loads.

• CPU Use: 1-2%

• RAM Use: Around 100MB.

• Windows 11

• i5-9300H

• 16GB RAM

[ItsPi3141]

Stuck Loading

The app gets stuck loading on any query. Never got past it.

OS

• OS: macOS Ventura 13.1 (intel)

Very strange because I tested it on an Intel MacOS virtual machine and it worked perfectly fine...

[ItsPi3141]

For those of you who say it's stuck on loading, can you check to make sure that the model is located on an SSD and not an HDD? And also make sure you have enough RAM.

[arjuna-dev]

My RAM is 32 GB 2667 MHz. I have the model in the computer's hard drive

[jamesblackwell]

Same issue, just stuck on loading. Mac m1, model is in my home dir.

I saw someone else recommend moving it to the models folder but I am not sure exactly where they mean

[rcooknet]

This screenshot is after 15 mins of runtime, still spinning. Macbook Pro M1 Max 32 GB ram, SSD, macOS 12.5 Monterey, v1.0.5 Alpaca Electron.

See no real memory usage at all from the Electron app in Activity Monitor, even thought the "RAM" graph on the Electron app shows nearly full usage.

[zvit]

Same. Haven't been able to use it yet since the day it came out - just spinning. I'm on SSD. 64GB RAM. 9% CPU in use. So many people have the same problem, it can't be a local issue. Please fix.

[Special-Niewbie]

I'm Testing the Software and I understand where is the problem here.

My CPU support AVX2 , it has NVMe and 32GB of RAM. The problem it seems the type of OS where you run the software, example it does't work on Windows 7 , previously I have used Windows 7 to test the software to have a lightweight OS, but after 2 Machine with similar specs that hang/stuck on loading I noticed that the software it doesn't use the CPU and RAM, so I moved to Windows 10 and 11 it does work properly and use the CPU and RAM and gives me the answers. I didn't test on my UBUNTU Linux yet, later I will try and let you know IF the software it will perform normally.

Edit: On Linux Ubuntu seems it works fine and use full of the hardware accesses and it takes more RAM (close to fill all 32GB of RAM) than the counterpart Windows.

[CRCODE22]

Same problem does not work here on Windows 11 32GB Ram and 3200MHZ AMD Ryzen 7 3700X 8 Cores 16 Logical Processes

Loads fully into RAM but after many minutes no output tried many different models 7B/13B.

[Supersonic734]

It stuck loading for me also

[genericgod]

Had the same problem. I launched the executable in Terminal.
Output was:

dyld: Symbol not found: __ZNSt3__113basic_filebufIcNS_11char_traitsIcEEE4openEPKcj
Referenced from: /Applications/Alpaca Electron.app/Contents/Resources/app/bin/chat_mac_x64 (which was built for Mac OS X 13.0)
Expected in: /usr/lib/libc++.1.dylib

My Mac is running on a lower MacOS than 13.
I then compiled llama.cpp on my machine and replaced /Alpaca Electron.app/Contents/Resources/app/bin/chat_mac_x64 with the new llama.cpp binary (main). Like it said in the README.

Now it works.

[jgersten23]

I have the same issue on a Macbook Air with 16gb of RAM which I know is pretty lowest-end for this kind of thing, but oddly I am able to run LLaMa in terminal using the same model (ggml-alpaca-7b-native-q4.bin) and it works fine and very quickly (although it hallucinates like a college junior in 1968).

Bit of an idiot here; I had been trying the non-ARM version (doh!) and when I removed it and installed the ARM version, it's now working perfectly and rapidly. Tremendous!

[carcinocron]

If there is an error message, is is possible to see what the error is or does this app silently discard it and keep showing the spinner?