Thank you for sending your enquiry! One of our team members will contact you shortly.
Thank you for sending your booking! One of our team members will contact you shortly.
Duration 7 hours
Course Outline
Transitioning from Text Generation to Typed Decisions
- Understanding the output format of a decision model.
- Contrasting system one decisions with reasoning LLMs.
- Analyzing the cost and latency characteristics of bounded decisions.
The Open-Weight Decision Model Ecosystem
- Strands Decider 2B, developed by AWS Labs.
- The JEV and GEV open-weight model families.
- Other models including Pplx-Decider, Liquid D1, and Clef.
- Differences in licensing and deployment strategies.
Anatomy of a Decision Model
- The LLM torso architecture, excluding the standard language modelling head.
- The pointer head mechanism used to score candidate options.
- The origin and calculation of the confidence score.
Executing Your First Decisions
- Setting up local environments on CPUs and consumer-grade GPUs.
- Formulating questions and defining option sets effectively.
- Processing multiple questions in a single pass.
- Configuring thresholds and fallback mechanisms.
Evaluating Decision Quality
- JevBench and other specialized decision benchmarks.
- Balancing accuracy with calibration.
- Interpreting Brier scores and the reliability of confidence metrics.
- Identifying scenarios where a decision model is an inappropriate choice.
Requirements
- A foundational understanding of large language models and prompt-based applications.
- Proficiency with command-line interfaces and Python scripting.
- No prior experience in machine learning training is necessary.
Target Audience
- AI engineers assessing decision models for use in agent workflows.
- Software developers looking to swap general LLM calls for bounded, specific decisions.
- Solution and platform architects building agent-based architectures.
- Technical product managers evaluating the viability of decision model options.
