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 structure of a decision model.
- Distinguishing system one decisions from reasoning LLMs.
- Analyzing the cost and latency profile of bounded decisions.
The Landscape of Open-Weight Decision Models
- Introduction to Strands Decider 2B from AWS Labs.
- Overview of the JEV and GEV open-weight families.
- Review of Pplx-Decider, Liquid D1, and Clef.
- Comparing licensing models and deployment variations.
Anatomy of a Decision Model
- Utilizing an LLM torso without a standard language modelling head.
- Scoring candidate options via a pointer head.
- Understanding the origin of confidence scores.
Executing Initial Decisions
- Setting up local environments on CPU and consumer-grade GPUs.
- Optimizing the phrasing of questions and option sets.
- Processing multiple queries in a single pass.
- Configuring thresholds and defining fallback behaviors.
Evaluating Decision Quality
- Using JevBench and other decision-specific benchmarks.
- Balancing accuracy against calibration.
- Interpreting Brier scores and the reliability of confidence metrics.
- Recognizing tasks where a decision model is not the appropriate tool.
Requirements
- A fundamental 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 required.
Target Audience
- AI engineers evaluating decision models for agent workflows.
- Software developers looking to replace LLM calls with bounded decision logic.
- Solution and platform architects designing agent-based systems.
- Technical product managers assessing various decision model options.
