Local Classifier

Classify text on your device by plugging an on-device classifier into Text Classifier or Sentiment Analysis, instead of a chat model.

What it does

Local Classifier is a model you plug into the Model slot of the Text Classifier or Sentiment Analysis node. With it connected, the classification runs in your browser: free, private and offline once the model is downloaded. What you connect decides how the node works; there is no switch.

It accepts two kinds of models:

• Zero-shot models (for example a DeBERTa zero-shot model) score the text against your own labels: the Categories of Text Classifier, or the Custom categories of Sentiment Analysis (positive, negative, neutral when empty).
• Fixed-label models (a sentiment or toxicity classifier) use their own labels. For Sentiment Analysis, common labels such as POSITIVE/NEGATIVE, LABEL_0 to LABEL_2, or 1 star to 5 stars are turned into positive, negative or neutral.

When to use it

• Tag or route text that must stay private or work offline.
• Classify a lot of text without paying per token.
• You need a label and a score, not an explanation.

To classify with reasoning or long instructions, connect a chat model such as Chat OpenAI instead.

Inputs and settings

Einstellung	Typ	Was es bewirkt
Model	Model picker	An on-device text classifier: zero-shot (uses your labels), sentiment or toxicity. Install one from Local AI, or download it from the picker.

The Model picker lists on-device classifiers: zero-shot, sentiment and toxicity models. Install one from the Local Models page, or download it from the picker. When you publish or share the workflow, Model is cleared, so whoever installs it picks their own.

Outputs

This node has no output port. It plugs into a slot on another node instead of passing items along.

• Model slot of Text Classifier, Sentiment Analysis

The connected node produces the output. With a Local Classifier, it also returns scores, a map of each label to its score (for example { "bug": 0.82, "question": 0.11 }), which you can use in a later IF node.

Dependencies and credentials

None. This node needs no connection and has no slots for other nodes.

Example workflow

1. A Context Menu trigger on selected text.
2. Text Classifier with Text {{ $input.selectionText }} and Categories billing, bug, feature request, other.
3. Local Classifier in its Model slot, with an installed zero-shot model.
4. Browser Notification with Message {{ $input.category }}.

Common issues

• “Choose an on-device classifier in the Local Classifier node.” Model is empty. Pick a model.
• “Model “…” isn’t available. Install it from Local AI.“ The model is not installed in this browser. Install it from Local Models, or download it from the picker.
• “Add at least one category: the zero-shot classifier scores the text against them.” Fill in Categories on the Text Classifier.
• My categories are ignored. The model is a fixed-label classifier (sentiment or toxicity). Use a zero-shot model to classify into your own labels.

Related nodes

• Text Classifier and Sentiment Analysis: the nodes that accept it.
• Classify Text: on-device classification as a standalone node.
• Local Q&A Model: the on-device model for the Q&A Agent.