Publish LumynaX Infused Gemma E4B Model

4d7be4d verified 5 months ago

2.98 kB

license: other
license_name: proprietary
license_link: >-
  https://huggingface.co/AbteeXAILab/lumynax-infused-gemma-e4b/resolve/main/LICENSE.txt
library_name: transformers
pipeline_tag: image-text-to-text
language:
  - en
  - mi
tags:
  - lumynax
  - lumynax-infused-gemma-e4b
  - gemma-4-e4b
  - multimodal
  - reasoning
  - audio
  - image
  - abteex-ai-labs
  - local-first

LumynaX Infused Gemma E4B Model

LumynaX Infused Gemma E4B Model is a real local release package built around the official google/gemma-4-E4B-it checkpoint, laid out in the same top-level structure as the checked-in Tiny and NZ 3B releases.

Provenance

upstream base model: google/gemma-4-E4B-it
package identity: LumynaX Infused Gemma E4B Model from AbteeX AI Labs
important limit: this is not a LumynaX fine-tuned checkpoint and not a LumynaX-merged Gemma checkpoint

Capabilities

reasoning mode: enabled
supported modalities from the upstream base model: text, image, audio
local runtime: Transformers-based package with the real upstream weights inside merged_model/
hardcoded package identity: merged_model/chat_template.jinja injects the LumynaX runtime identity by default

Included Local Structure

merged_model/: real packaged upstream Gemma 4 E4B model files
merged_model/LUMYNAX_PACKAGE_IDENTITY.txt: packaged identity prompt for the LumynaX runtime wrapper
artifacts/release_training_summary.json: release metadata and provenance summary
hf_space/: private reasoning + multimodal Gradio demo aligned to this release package
ollama/: included for parity with other releases, but not validated for this multimodal base package
release_export_manifest.json: local manifest for this package

Important Limit

This package carries the real upstream Gemma 4 E4B weights, but it does not claim any LumynaX-specific adaptation of those weights. It should be treated as a locally packaged LumynaX-branded Gemma base release under the LumynaX Infused Gemma E4B Model release identity, not as a true LumynaX-infused checkpoint.

Quick Start

pip install -r requirements.txt
python quickstart.py
python quickstart.py --mode image --image path-or-url
python quickstart.py --mode audio --audio path-or-url

The default python quickstart.py path exercises a local text + reasoning flow.

Consumer Runtimes

Ollama

The package keeps an ollama/ folder so the release shape stays close to the existing Tiny and NZ 3B exports. That path has not been validated for this multimodal base package.

ollama create lumynax-infused-gemma-e4b -f ollama/Modelfile
ollama run lumynax-infused-gemma-e4b

Zero-Install Browser Demo

The included hf_space/ bundle is a private reasoning + multimodal Gradio demo for this package. Recommended default Hugging Face model repo id:

AbteeXAILab/lumynax-infused-gemma-e4b