ThoxWebby-Gemma-4-E2B

Gemma-4-E2B-it-qat-mobile-transformers derivative — the browser-tier ThoxOS build: a Gemma-4-E2B instruction-tuned, quantization-aware mobile checkpoint intended to run through Transformers.js on WebGPU; this repository currently carries th

google/gemma-4-E2B-it-qat-mobile-transformers

Find your model: THOX Model Factory · Collection: iOS, Browser & Device Knowledge Packs

Your AI. Your Data. Your Rules.™

Status: WEIGHTS NOT ATTACHED. This repository currently carries the model card and packaging; the weights are not attached here yet. See Specifications for exactly what this repo contains.

Summary

The browser-tier ThoxOS build: a Gemma-4-E2B instruction-tuned, quantization-aware mobile checkpoint intended to run through Transformers.js on WebGPU; this repository currently carries the model card only, with no weight files.

Tags: thox · thoxos · meshstack · browser · webgpu · qat · gemma

Highlights

• The browser-tier ThoxOS build: a Gemma-4-E2B instruction-tuned, quantization-aware mobile checkpoint intended to run through Transformers.js on WebGPU; this repository currently carries the model card only, with no weight files.
• There is nothing to install from this repository: it contains this card and no model files.

Specifications

Base model	google/gemma-4-E2B-it-qat-mobile-transformers
Parameters	E2B
License	The upstream Gemma terms govern the base model (see the license link in this card's metadata and the disclosure in the sections above); this repository distributes no weights.
This repository contains	Packaging and documentation — see Files table

Files

File	Size	Purpose
.gitattributes	1,519 B	Repository file

At a glance

Field	Value
Base model	google/gemma-4-E2B-it-qat-mobile-transformers
Fine-tune	Not recorded in this repo — no adapter, training log, or dataset statement is published
Architecture	Gemma 4 E2B instruction-tuned, quantization-aware mobile checkpoint; layer and head counts are not stated here
Parameters	Not published in this repo — no weight files are present
Quantization	QAT mobile build, as named in the base model id; no artifacts are published here
Native context	Not published in this repo
Prompt template	Not published in this repo
License	Gemma Terms of Use (declared here as gemma with license_link to the Gemma 4 licence) — access public
Publisher	THOX.AI LLC

Quick start

There is nothing to install from this repository: it contains this card and no model files. The intended deployment is Transformers.js in a browser, loading weights published under an approved path for the ThoxOS browser tier:

import { pipeline } from "@huggingface/transformers";
const pipe = await pipeline("text-generation", "<weights-path>", { device: "webgpu" });

Access is public and ungated (ungated 2026-10-31 on the CTO's standing directive: no gating unless licensing requires it). This repository distributes no weights; the licence declared in this card is the one recorded on the repository, and the linked Gemma 4 licence governs the upstream model.

Platforms

Platform / device	Run path
Hugging Face	Download the builds listed under Files from this repo (see Run it)

Intended use

On-device chat and short task replies inside the ThoxOS browser tier, running locally on the user's GPU through WebGPU, fully offline once the weights are cached, with facts supplied by the serving layer.

Out of scope: safety-critical, medical, or legal decisions without a human in the loop; server-side deployment as a general assistant; any use as a standalone source of THOX facts.

Training data and procedure

THOX fine-tune of google/gemma-4-E2B-it-qat-mobile-transformers. The corpus and procedure are described in this card as published at release (see Summary and Evaluation); no training beyond what this card describes is claimed for this repository.

Evaluation

No evaluation artifact is published for this repository and no figure is verified; this repository carries the model card only and distributes no weights (see Limitations).

Limitations

• No weights, tokenizer, or ONNX export is published in this repository, so nothing here is runnable as shipped.
• Model size, context length, and quality are not recorded here; no evaluation artifact is published, so no figure is verified.
• Browser memory is the binding constraint for this tier: the runtime must fit the WebGPU device's memory and be prepared to fall back to WASM.
• The licence declared on the repository (gemma) and the linked upstream Gemma terms must both be checked before redistribution; where they conflict, the upstream Gemma terms govern.
• Behaviour and bias reflect the upstream checkpoint; the THOX fine-tune is not documented here.

Links

• Model: Thox-ai/ThoxWebby-Gemma-4-E2B
• Base model: google/gemma-4-E2B-it-qat-mobile-transformers
• Upstream licence: Gemma 4 licence
• Organisation: huggingface.co/Thox-ai

---

Published by THOX.AI LLC.

Provenance

Published by THOX.ai as Thox-ai/ThoxWebby-Gemma-4-E2B. Upstream base: google/gemma-4-E2B-it-qat-mobile-transformers. This card was conformed to THOX Model Card Standard v1.0 on 2026-10-31 (frontmatter, tags, header, files, and section structure); the technical content is as published at release.

License

The upstream Gemma terms govern the base model (see the license link in this card's metadata and the disclosure in the sections above); this repository distributes no weights.

Citation

@misc{thoxai2026thoxwebbygemma4e2b,
      title={ThoxWebby-Gemma-4-E2B},
      author={THOX.ai},
      year={2026},
      url={https://huggingface.co/Thox-ai/ThoxWebby-Gemma-4-E2B}
}