How deeply does Building LLM Applications with DSPy cover single-label categorical classification?

Defines label cardinality, gives a multi-intent counterexample, and notes that the simpler setup can be too limiting.

Single-label categorical classification

As explained in Building LLM Applications with DSPy

A single-label problem is a classification setup in which each input is assigned at most one class label.

In single-label classification, each input is assigned at most one class. This is separate from the binary-versus-multiclass distinction, which depends on the number of possible classes. section 3.1.2

Intent classification

An utterance asking for a flight and food options contains two intents, so it does not fit a single-label setup. In the ATIS data, combined-intent classes are removed to simplify classification. section 3.1.2 section 3.2

Listing 3.3 Mapping the intents to their new names

df['intent'] = df['intent'].map(ATIS_INTENT_MAPPING)  #1
df = df.dropna(subset=['intent'])  #2
print(f"DF number of rows after removing multi label classes: {len(df)}")

The mapping and row filtering show how combined-intent classes are excluded when preparing the ATIS labels. section 3.2

How Building LLM Applications with DSPy covers it

Cover of Building LLM Applications with DSPy

liveBook manning.com

Read Building LLM Applications with DSPy on liveBook Get the book at manning.com

Also explained in

liveBook manning.com

Deep Learning with Python, Third EditionA classification task in which each example belongs to exactly one class from a set of mutually exclusive classes.

Compare how 2 books explain single-label categorical classification →

Related concepts

multilabel classificationcontrasts with · Multilabel classification is a classification task in which each input may be assigned zero or more labels simultaneously.

This explainer was generated by Manning's knowledge graph from the text of Building LLM Applications with DSPy. The section links point to the parts of the book it draws on.