cgg 0.3.9

cgg is a short form of call gguf model/file; cgg is a cmd-based app built on gguf-connector, which allows users interacting with large language model (i.e., chatgpt) via a simple command without coding a long long syntax

GGUF caller

This package is a GGUF (GPT-Generated Unified Format) file/model caller. Perfect to run in low end device(s).

install the caller via pip/pip3 (once only):

pip install cgg

If your c/c++ compiler cannot build the engine wheel successfully; could opt to get it here straight; or try core.

update the caller (if not in the latest version) by:

pip install cgg --upgrade

user manual

This is a cmd-based (command line) package, you can find the user manual by adding the flag -h or --help.

cgg -h

check current version

cgg -v

connector menu

cgg menu

graphical user interface (GUI) caller

Simple GUI to interact with a chat model for generating responses (recommended).

call cpp connector:

cgg cpp

call c connector:

cgg c

command line interface (CLI) caller

CLI input also provided for expert/advanced user(s).

call gpp connector:

cgg gpp

call g connector:

cgg g

call v connector:

cgg v

GGUF file(s) in the same directory will automatically be detected by the caller, hence, a selection menu will be shown in the console as below.

metadata reader:

You can check model metadata by:

cgg r

Or try the model analyzor by:

cgg a

clone feature

cgg clone [url]

With this fast clone feature, you can clone any (GGUF model) file from URL, save it automatically in the current directory, and get it ready to connect locally (run it with your own machine offline); depends on the file size, as well as the network connectivity, it may take a while to complete the clone process; and you are able to see a dynamic progress bar showing the downloading status.

If you are a mac newbie, you might encounter ssl cert. issue while executing clone command or while entering the lazylist below; a practical solution: click Install Certificates.command under your Python version folder; details please refer to the issues reported.

sample model(s) available to download (try out)

For general purpose [chat.gguf] (size: around 2GB or less)

cgg clone https://huggingface.co/calcuis/chat/resolve/main/chat.gguf

For coding assistance [code.gguf] (size: around 3GB or more)

cgg clone https://huggingface.co/calcuis/code_mini/resolve/main/code.gguf

For health/medical advice [medi.gguf] (size: around 3GB or more)

cgg clone https://huggingface.co/calcuis/medi_mini/resolve/main/medi.gguf

For law/legal opinion [law.gguf] (size: around 3GB or more)

cgg clone https://huggingface.co/calcuis/law_mini/resolve/main/law.gguf

For wealth/financial plan [finance.gguf] (size: around 3GB or more)

cgg clone https://huggingface.co/calcuis/gguf/resolve/main/finance.gguf

For calculation/math solution [math.gguf] (size: around 3GB or more)

cgg clone https://huggingface.co/calcuis/gguf/resolve/main/math.gguf

***those are all experimental models; no guarantee on quality

sample model list

You can either use the clone feature above or opt it straight from the lazy list by:

cgg s

comfy gguf pack

Download comfy GGUF windows portable pack by:

cgg y

pdf analyzor (beta)

You can load PDF into the model for analysis right away by (analyzor c):

cgg pc

Or by (analyzor p):

cgg pp

voice/speech recognizor (beta)

You can prompt voice/speech (wav file) as input by (recognizor c):

cgg vc

Or by (recognizor p):

cgg vp

Or by (online recognizor c):

cgg oc

Or by (online recognizor p):

cgg op

website

Click gguf.io (mirror of gguf.us) or launch it straight from console by:

cgg io

Click gguf.us or launch it straight from console by:

cgg us