ARTLAS-HUB-LLM / README.md
mfiguero2's picture
AR.T.L.A.S. HUB LLM: ficha y catálogo provisorio (24 modelos)
1881606 verified
|
Raw History Blame Contribute Delete
3.12 kB
---
pretty_name: AR.T.L.A.S. HUB LLM
language:
- es
- en
tags:
- gguf
- llm
- catalog
- local-llm
- artlas
size_categories:
- n<1K
configs:
- config_name: catalogo
data_files: hub.json
---
# AR.T.L.A.S. HUB LLM
*Español · [English](#english)*
Biblioteca de modelos de lenguaje **gratuitos y de pesos abiertos** del ecosistema AR.T.L.A.S.
(*Argentine Transformer for Learning and Advanced Semantics*), de **AI-TECH** — [ai-tech.ar](https://ai-tech.ar).
Este repositorio publica el catálogo `hub.json`, que se actualiza todos los días. Cada ficha indica
dónde está el modelo, qué licencia tiene, cuánto pesa, qué equipo necesita y cómo rinde sin GPU.
## Qué contiene `hub.json`
| Campo | Significado |
| --- | --- |
| `id`, `nombre`, `familia`, `autor` | Identificación del modelo |
| `repo`, `url` | Repositorio de origen en Hugging Face |
| `licencia`, `licencia_libre` | Licencia declarada y si es libre (Apache-2.0, MIT, BSD, CC-BY-4.0, CC0) |
| `nivel`, `ram_min_gb`, `cpu` | Nivel de hardware, RAM mínima y rendimiento estimado sin GPU |
| `recomendado`, `variantes` | Cuantización sugerida y archivos GGUF con tamaño y sha256 |
| `capacidades` | texto, visión, herramientas, razonamiento, código |
## Dónde están los modelos
Los pesos permanecen en los repositorios de sus autores; aquí no se redistribuyen.
Cada modelo conserva su propia licencia: revísela antes de usarlo, en especial los marcados
con `licencia_libre: false`.
## Cómo se usa
Desde AR.T.L.A.S. Nexus, AR.T.L.A.S. OS y AR.T.L.A.S. RAG, que leen este catálogo y descargan
el archivo GGUF elegido directamente al equipo del usuario.
Contacto: contacto@ai-tech.ar
---
## English
**AR.T.L.A.S. HUB LLM** is the library of **free, open-weight** language models of the AR.T.L.A.S. ecosystem
(*Argentine Transformer for Learning and Advanced Semantics*), by **AI-TECH** — [ai-tech.ar](https://ai-tech.ar).
This repository publishes the `hub.json` catalog, updated daily. Each entry says where the model lives, its license,
its size, the hardware it needs and how it performs without a GPU. Level names (`nombre_en`) and notes
(`nota_en`, `licencia_nota_en`) are provided in English as well.
| Field | Meaning |
| --- | --- |
| `id`, `nombre`, `familia`, `autor` | Model identification |
| `repo`, `url` | Source repository on Hugging Face |
| `licencia`, `licencia_libre` | Declared license and whether it is free (Apache-2.0, MIT, BSD, CC-BY-4.0, CC0) |
| `nivel`, `ram_min_gb`, `cpu` | Hardware tier, minimum RAM and estimated performance without a GPU |
| `recomendado`, `variantes` | Suggested quantization and GGUF files with size and sha256 |
| `capacidades` | text, vision, tools, reasoning, code |
**Where the models live.** Weights stay in their authors' repositories; they are not redistributed here.
Each model keeps its own license: review it before use, especially entries with `licencia_libre: false`.
**How it is used.** AR.T.L.A.S. Nexus, AR.T.L.A.S. OS and AR.T.L.A.S. RAG read this catalog and download the chosen
GGUF file straight to the user's machine.
Contact: contacto@ai-tech.ar