Thox-ai

ThoxClip-Qwen3-0.6B

README

License: apache-2.0

Summary

ThoxClip-Qwen3-0.6B is the published on-device assistant for ThoxClip: a LoRA fine-tune (rank 32) on a THOX-authored corpus of 205 chat and tool-use examples, merged into the base weights. Status: PUBLISHED, grounding required. It must be served with the grounded THOX device-facts system prompt; bare deployment is out of contract.

Highlights

THOX-authored fine-tune: LoRA (rank 32) on a 205-example THOX corpus of chat and tool-use behaviour, merged into the base weights.
Known-false discipline at gate: 9/10 on the GATE49 known-false category, the comparatively strong axis of this baseline.
Published as the ThoxClip baseline: serviceable under grounding with MeshStack escalation for questions outside the trained facts.

Specifications

Table

Base model	Qwen/Qwen3-0.6B (Apache-2.0)
Parameters	0.6B
Formats	Merged safetensors in this repo · GGUF served via Ollama
Training	LoRA rank 32 on a 205-example THOX-authored corpus, merged into the base
License	Apache-2.0. The THOX fine-tune and additions in this repo are covered under Apache-2.0; see the license text for terms.
This repository contains	The merged full weights, tokenizer, chat template, and training summary

Files

Table with columns: File, Size, Purpose
File	Size	Purpose
`model.safetensors`	1,192,135,096 B	Model weights (safetensors)
`tokenizer.json`	11,422,650 B	Tokenizer files
`chat_template.jinja`	4,168 B	Chat template
`.gitattributes`	1,570 B	Repository file
`config.json`	1,420 B	Model configuration
`tokenizer_config.json`	694 B	Tokenizer files
`training_summary.json`	430 B	JSON metadata
`generation_config.json`	214 B	Generation configuration

Run it

Full weights (safetensors) are in this repo:

python

from transformers import AutoModelForCausalLM, AutoTokenizermodel = AutoModelForCausalLM.from_pretrained("Thox-ai/ThoxClip-Qwen3-0.6B")

The GGUF build is served via Ollama:

bash

ollama pull Thox-ai/thoxclip-qwen3-0.6bollama run Thox-ai/thoxclip-qwen3-0.6b

Serve thinking-off, with the grounding system prompt applied at the serving layer.

Platforms

Table with columns: Platform / device, Run path
Platform / device	Run path
Hugging Face	Full weights (safetensors) in this repo — `AutoModelForCausalLM.from_pretrained` (see Run it)
Ollama	`ollama run Thox-ai/thoxclip-qwen3-0.6b` — serve thinking-off with the grounding system prompt

Intended use

On-device assistant for ThoxClip: THOX product questions, device help, and small-model chat in the THOX house voice, always under the grounding prompt.
Out of scope: thinking-mode serving, long-form generation, bare (ungrounded) deployment, safety-critical, medical, or legal decisions, and any THOX claim beyond the published record.

Training data and procedure

THOX fine-tune of Qwen/Qwen3-0.6B. 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

Measured with the THOX GATE49 factuality gate (49 questions, weights only, no serving-side grounding; GATE49 is a development set, abstention-weighted) at this build's finishing gate. Category results as reported at release.

Table with columns: Benchmark, Result
Benchmark	Result
Verified-fact recall	2/12
Known-false	9/10
Pricing	0/1
Hard failures	14
Total	22/49 — mean 2.776/5

THOX GATE49 factuality gate (49 questions, weights only, no serving-side grounding), measured at this build's finishing gate: 22/49 passed, mean 2.776/5, 14 hard failures. Verified-fact recall is weak (2/12), the pricing question is not declined (0/1), and the known-false category is comparatively strong (9/10). This is not a strong gate result and THOX does not present it as one; the build is published as the ThoxClip baseline because its behaviour is serviceable under grounding and mesh escalation, with the grounding requirement stated plainly.

Limitations

Grounding required: factual reliability depends on the grounding system prompt and on MeshStack escalation for questions outside the trained facts.
Weak verified recall at this size; expect correct behaviour mainly on trained facts and declines.
Thinking-mode behaviour is out of contract; serve thinking-off.
This build has not been qualified on ThoxClip device hardware.
The fine-tune corpus is small (205 examples). It teaches THOX identity, house voice, and tool-routing behaviour; it does not add general capability.

Provenance

Published by THOX.ai as Thox-ai/ThoxClip-Qwen3-0.6B. Upstream base: Qwen/Qwen3-0.6B. 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

Apache-2.0. The THOX fine-tune and additions in this repo are covered under Apache-2.0; see the license text for terms.

Back THOX on Kickstarter

THOX.ai is live on Kickstarter. Explore ThoxKey, ThoxAir, ThoxMini and ThoxClip, and help bring the THOX personal AI ecosystem to life.

Back THOX on Kickstarter

Citation

bibtex

@misc{thoxai2026thoxclipqwen306b,      title={ThoxClip-Qwen3-0.6B},      author={THOX.ai},      year={2026},      url={https://huggingface.co/Thox-ai/ThoxClip-Qwen3-0.6B}}

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Model Details

Model Provider