| --- |
| license: apache-2.0 |
| tags: |
| - intent-classification |
| - text-classification |
| - logistic-regression |
| - sklearn |
| - conference-talk-demo |
| language: |
| - en |
| library_name: sklearn |
| --- |
| |
| **⚠️ Conference talk demo — not production weights.** |
|
|
| This model accompanies a conference keynote on local on-device AI. Published as a reference for the fine-tuning patterns shown on stage — **not** a deployable artefact. No security audit, no SLA, pinned to the talk's state. |
|
|
| - Source repository: [thinktecture-labs/local-multi-model-agent-slm](https://github.com/thinktecture-labs/local-multi-model-agent-slm) |
| - Threat model + out-of-scope: [SECURITY.md](https://github.com/thinktecture-labs/local-multi-model-agent-slm/blob/main/SECURITY.md) |
| - Licensing details: [MODEL_LICENSES.md](https://github.com/thinktecture-labs/local-multi-model-agent-slm/blob/main/finetune/MODEL_LICENSES.md) |
| - All five models in the stack: [Collection — Local Multi-Model Agent — nextera fine-tunes](https://huggingface.co/collections/thinktecture/local-multi-model-agent-nextera-fine-tunes-6a04a8ff2a40e5696f3c2f18) |
|
|
| --- |
|
|
| ## LogReg Intent Classifier |
|
|
| | | | |
| |---|---| |
| | **Base** | scikit-learn `LogisticRegression`, multinomial, L2 penalty | |
| | **License** | Apache-2.0 (this repo) — but inputs are EmbeddingGemma vectors so the [Gemma Terms](MODEL_LICENSES.md) cover the embedding step | |
| | **Training script** | [`training/train_intent_logreg.py`](../training/train_intent_logreg.py) | |
| | **Method** | LogReg on FT-EmbeddingGemma's 768-dim output vectors. Held-out 90/10 split. ~2 minutes on CPU. | |
| | **Training data** | Same as Gemma3-1B intent: `data/training-data/gemma3_intent_{scenario}.jsonl` (re-embedded with the FT EmbeddingGemma) | |
| | **Hardware** | CPU is sufficient. Requires the FT EmbeddingGemma llama-server running on port 9092/9096 to embed training examples. | |
| | **Intended use** | Replaces the 1B generative classifier as the primary intent router. ~10ms per query (vs ~200ms for the 1B). Same accuracy on the standard eval set. | |
| | **Out of scope** | Anything that requires generation (it's a 3-way classifier). Low-confidence predictions (< 0.60 threshold, configurable in `intent_classifier_logreg.py`) are overridden to `direct_answer` as a safe fallback intent. The 1B generative classifier is only used as a load-time fallback when the LogReg model file is absent, not as a per-query confidence fallback. | |
| | **Reference eval (Nextera)** | 96.1% on 180-query eval set. ~10ms per classification (single CPU thread). | |
| | **Known failure modes** | When the EmbeddingGemma FT changes, the LogReg weights become invalid — `intent_classifier_logreg.py:13-15` warns about this coupling. Re-train both together. | |
|
|
