Pre-trained Classifiers
We offer the following pre-trained classifiers which you can use right away to classify text data. To use one, follow the instructions in the Quickstart and replace the "model" parameter with the model ID from this table.
| Name | Model ID | Description |
|---|---|---|
| ONET-SOC | job_postings_onet | Classify job descriptions/postings into 900+ ONET-SOC job categories. |
| IAB Products | iab_products | Classify product listings or reviews into 500+ IAB product categories. |
| IAB Content | iab_content | Classify web content into 700+ IAB content categories. |
| NAICS Industries | naics | Classify business descriptions into 100+ NAICS industry categories. |
| MCC Categories | mcc | Classify businesses/merchants into 100+ MCC merchant categories. |
| Topics | topics | Classify text into 1,000+ hierarchical topics. |
| Intents | intents | Classify user queries into 100+ intents. |
| IPTC Media Topics | iptc_media_topics | Tag content with over 1000 common media topics. |
| GARM Brand Safety | garm_brand_safety | Flag content with potential brand safety concerns. |
Testing the Pretrained Classifiers
To quickly test the classifiers, you can sign in and use the models via the Playground. The Playground allows you to test the models, configure the max labels and threshold, and validate the results in real-time. When you're ready to integrate the model in your production environment, you can copy the request code to make requests to the model.