cnnmmd

cnnmmd_xoxxox_tttlcp

Engines: AI Related: Language Generation Engine: Local (Gemma 3n [CPU/GPU])

Implementation/Rights (Source Code/License): [※E]

・

https://github.com/cnnmmd/cnnmmd_xoxxox_tttlcp

※E

This code is available to the public so you can see how it works (it's still being cleaned up for collaboration).

dependence:

・

cnnmmd_xoxxox_libtlk

overview

This is a language generation environment and model (llama.cpp + quantized model) that can be run on a general-purpose PC (CPU only): [※1]

・

https://github.com/ggml-org/llama.cpp

・

https://huggingface.co/ggml-org/gemma-3n-E2B-it-GGUF

・

Operating environment: Local (CPU/GPU)

*1

Gemma 3n is multimodal, but this version is quantized to run in the smallest environments and currently does not include image recognition.

Installation

This plugin creates the following resources: [*1] [*2]

・

Container: xoxxox_envlcp: 00.7GB

・

Folder: cnnmmd/import/cnnmmd_xoxxox_ttslcp/applcp: 05.0GB

*1

The container will work with 4GB of memory (RAM), but it will take several minutes to start the model (and subsequent conversations will tend to be slow). With 6GB - 16GB, the response will be reasonably fast.

*2

Since it runs on a container, Metal is not available even on Mac.