Purebasic and chat gpt (version 3.5)

[idle]

I've re uploaded it, I had missed out one of the MinGW runtime dlls

@Jak64, Yes it runs locally on your desktop, the falcon model seems to be the quickest, though it depends what you want to do.

[jak64]

hello Caronte3D

1) I downloaded "https://dnscope.io/idlefiles/gpt4all_pb.zip" and unzipped it.

2) I clicked on the link "https://gpt4all.io/index.html" and on the displayed page, I clicked on "Windows installer"

3) I installed the program in the default directory "C:\Users\DELL\gpt4all".

I don't know what to do next.

Can you help me ?

THANKS

[idle]

hello Caronte3D

1) I downloaded "https://dnscope.io/idlefiles/gpt4all_pb.zip" and unzipped it.

2) I clicked on the link "https://gpt4all.io/index.html" and on the displayed page, I clicked on "Windows installer"

3) I installed the program in the default directory "C:\Users\DELL\gpt4all".

I don't know what to do next.

Can you help me ?

THANKS

you need to "https://gpt4all.io/index.html" and half way down the page to where it says "Model Explorer"
and download a model like "ggml-model-gpt4all-falcon-q4_0.bin"
then save it to the same folder you extracted gpt4all_pb

[jak64]

Hui Caronte3D,

1) I copied the model "ggml-model-gpt4all-falcon-q4_0.bin" to the directory "gpt4all_pb\bin\

2) I modified line 277 as follows:

path.s = GetPathPart(ProgramFilename()) + "ggml-model-gpt4all-falcon-q4_0.bin"

3 I launched the program "gpt4all.pbi"

I get the following error message:

llama.cpp: loading model from D:\Purebasic\ChatGPT\gpt4all_pb\bin\ggml-model-gpt4all-falcon-q4_0.bin
model loading error: missing tok_embeddings.weight
llama_init_from_file: Failed to load model
LAMA ERROR: Failed to load model from D:\Purebasic\ChatGPT\gpt4all_pb\bin\ggml-model-gpt4all-falcon-q4_0.bin

[idle]

you might need to look at compiler options and check that your creating temporary executable in the source directory.

[jak64]

Hello Caronte3D,

1) I haven't changed anything in the compilation options (I don't know what should be changed)

2) I compiled the program and created an executable and upon running the executable I get the same error

[idle]

And was the exe in the bin folder.

[Kurzer]

Runs like a charme here.

jak64: Did you downloaded the correct model file? Filesize of 'ggml-model-gpt4all-falcon-q4_0' should be 4.061.641.216 Bytes.

PS: I notice that the PB program now doesn't give any status message at all when the model loads successfully (also no memory usage etc). Have you removed this, Idle?

[idle]

Umm no it should still be printing the status to console. I'm away from the PC but will look in the morning and make a separate thread about it tomorrow.

[Caronte3D]

1) I copied the model "ggml-model-gpt4all-falcon-q4_0.bin" to the directory "gpt4all_pb\bin\
gpt4all_pb\bin\ggml-model-gpt4all-falcon-q4_0.bin

Put *ALL* the files in the same directory (not \bin) where you compile your .exe and try again.

[jak64]

Hello Idle, Kurzer, Caronte3D,

Thank you for your messages.

1) The .exe was in the bin directory.

2) The "ggml-model-gpt4all-falcon-q4_0.bin" file weighs 3,966,447 bytes.

3) I copied all the files directly under the "\gpt4all_pb" directory

I recompiled, created the .ex and launched the program, I have the same error message.

My computer has 16 GB of RAM.

[Caronte3D]

Try another model.
I used this without problem: nous-hermes-13b.ggmlv3.q4_0.bin

[jak64]

Hello Caronte3D,
It's 1:00 am, at home, and I'm only answering now because I live in Ciboure in France, a beautiful little town on the edge of the Atlantic Ocean and my sister and my brother-in-law have come to spend holidays with me .

I'll try with another model (if I know how to download it!!!) and I'll let you know.

I would like to thank very warmly all of you who help me because when I manage to make all this work, it will be extraordinary for me for the use I want to make of it.

See you

ps: I use google to do the English translation, hence perhaps not very correct sentences!!!

[idle]

this is the list of files, the paths shouldn't matter as long as they're all in the same directory.

Code: Select all

"G:\gpt4all\gpt4all-backend\build\bin\libllmodel.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libllmodel.lib"
"G:\gpt4all\gpt4all-backend\build\bin\libmpt-avxonly.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libmpt-default.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libreplit-mainline-avxonly.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libreplit-mainline-default.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libstarcoder-avxonly.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libstarcoder-default.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libunwind.dll"
"G:\gpt4all\gpt4all-backend\build\bin\ggml-model-gpt4all-falcon-q4_0.bin"    ;<- the model you downloaded 
"G:\gpt4all\gpt4all-backend\build\bin\ggml-replit-code-v1-3b.bin"
"G:\gpt4all\gpt4all-backend\build\bin\gpt4all.pbi"                                                    
"G:\gpt4all\gpt4all-backend\build\bin\gpt4all.pbi.cfg"
"G:\gpt4all\gpt4all-backend\build\bin\libbert-avxonly.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libbert-default.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libc++.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libfalcon-avxonly.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libfalcon-default.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libgptj-avxonly.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libgptj-default.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libllamamodel-230511-avxonly.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libllamamodel-230511-default.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libllamamodel-230519-avxonly.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libllamamodel-230519-default.dll"
"G:\gpt4all\gpt4all-backend\build\bin\libllamamodel-mainline-avxonly.dll"
"G:\gpt4all\gpt4all-backend\build\bin\PureBasic_Compilation0.exe"             ;<-the temporary exe generated when hitting F5 

[jak64]

Hello idle,
Thank you for your answer.
I'll watch it tomorrow (it's 1:35 am in France and I'm going to bed).
Good day to you.
See you