DistilBERT Customer Support Intent Classifier

Fine-tuned DistilBERT model for 27-class customer support intent detection.

Evaluation Results

Metric Score
Test Accuracy 0.9981
Test Macro F1 0.9981
Test Weighted F1 0.9981

Usage

from transformers import pipeline

classifier = pipeline(
    "text-classification",
    model="Kaave19/distilbert-customer-support-intent"
)

classifier("I want to cancel my order from yesterday")

## Limitations

This model is intended for customer-support-style messages similar to the Bitext dataset. It predicts one intent per message.
Downloads last month
3
Safetensors
Model size
67M params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for Kaave19/distilbert-customer-support-intent

Finetuned
(11861)
this model

Dataset used to train Kaave19/distilbert-customer-support-intent