GPT-4

2 use cases using this technology

2published use cases

2industries

0countries on record

AI Model Development & MLOps

top AI capability

Evidence mix: High 2 · Medium 0 · Low 0 — bands are computed from each record's evidence signals.

Industry

2 use cases

Technology & SoftwareGenerative AI

Kantar Worldpanel fine-tunes GenAI models on Databricks to generate market-insight training data faster

Kantar Worldpanel used the Databricks Data + AI Platform and MLflow to experiment with Llama, Mistral, GPT-4 and GPT-3.5 for a proof of concept linking receipt descriptions to product barcode names. GPT-4 produced the most accurate outputs (94%), which the team used to automatically generate a training dataset of about 120,000 receipt-to-barcode pairs in a couple of hours to fine-tune a smaller production model.

Kantar WorldpanelDatabricks · MLflow · Unity Catalog +3

Human ResourcesLarge Language ModelsAI Model Development & MLOpsMachine Learning

Checkr: Streamlining Background Check Classification with Fine-tuned Small Language Models

Checkr automated adjudication of background checks using a tuned logistic regression model for 98% of cases, but the remaining 2% required classifying noisy records into 230 categories. After testing GPT-4 (80-82% accuracy on hard cases, ~15s latency, ~$12k cost) and GPT-4 with RAG, Checkr fine-tuned Llama-2-7b (85% accuracy) and then, in production on Predibase, Llama-3-8b-instruct, achieving 90% accuracy on the hardest 2% of cases with 0.15-second response times (30x faster than GPT-4) and a 5x cost reduction, using LoRA/LoRAX for parameter-efficient fine-tuning and multi-adapter serving.

CheckrLlama-2-7b · Llama-3-8b-instruct · LoRA +3