ThoxNano-135M

ThoxKey Tier Model — the smallest THOX device model, built for the ThoxKey tier: a LoRA fine-tune (rank 32) on a THOX-authored corpus of 982 chat, tool-use, and factuality examples, merged into the base weights

GATE49 23/49 (development set) · Safetensors weights · GGUF: Q4_K_M · Live on Ollama

Your AI. Your Data. Your Rules.™

Status: PUBLISHED — GROUNDING REQUIRED. Deployed surfaces supply facts through the serving layer; the bare weights are not a reliable source of THOX facts. The full gate result is printed under Evaluation.

Summary

ThoxNano-135M is the smallest THOX device model, built for the ThoxKey tier: a LoRA fine-tune (rank 32) on a THOX-authored corpus of 982 chat, tool-use, and factuality examples, merged into the base weights. Status: PUBLISHED, grounding required. The Key tier is mesh-served: a connected host computer does the compute, and facts are supplied by the serving layer.

Highlights

• THOX fine-tune: a LoRA fine-tune (rank 32) on a THOX-authored corpus of 982 chat, tool-use, and factuality examples, merged into the base weights.
• GATE49 23/49 on the THOX factuality gate (development set, abstention-weighted), mean 3.0/5, 10 hard failure(s) — printed in full under Evaluation.
• This repo carries the merged weights (safetensors), a Q4_K_M GGUF, an Ollama Modelfile, and the chat template (chat_template.jinja, shipped in this repo).

Specifications

Files

Run it

This repo carries the merged weights (safetensors), a Q4_K_M GGUF, an Ollama Modelfile, and the chat template (chat_template.jinja, shipped in this repo).

ollama pull Thox-ai/thoxnano-135m
ollama run Thox-ai/thoxnano-135m

hf download Thox-ai/ThoxNano-135M ThoxNano-135M.Q4_K_M.gguf --local-dir ./ThoxNano-135M

Platforms

Intended use

• Very short conversational replies and routing intents for the ThoxKey tier, with facts supplied by the serving layer and a larger mesh node available for anything beyond trivial exchanges.
• Out of scope: safety-critical, medical, or legal decisions without a human in the loop, and any use as a standalone source of THOX facts.

Training data and procedure

THOX fine-tune of HuggingFaceTB/SmolLM-135M. 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). Full narrative under Evaluation below.

THOX GATE49 factuality gate (49 questions, weights only, no serving-side grounding), most recent evaluation of the shipped build: 23/49 passed, mean 3.0/5, 10 hard failures. This is a weak gate result, the expected posture at this size; the model is published for identity, voice, and routing behaviour under grounding, not for factual recall from the weights.

Limitations

• At 135M parameters the capability ceiling is very low: expect short, shallow answers and occasional template echoes.
• Factual answers from the bare weights are unreliable; deploy with serving-side grounding for facts.
• This build has not been qualified on ThoxKey device hardware.

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.

THOX.ai — Your AI. Your Data. Your Rules.™

Provenance

Published by THOX.ai as Thox-ai/ThoxNano-135M. Upstream base: HuggingFaceTB/SmolLM-135M. 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.

Citation

@misc{thoxai2026thoxnano135m,
      title={ThoxNano-135M},
      author={THOX.ai},
      year={2026},
      url={https://huggingface.co/Thox-ai/ThoxNano-135M}
}