Reasoning loops that observe, plan, call tools, verify results, and maintain useful state over a task.
Interleaving reasoning with tool observations lets an agent adapt its plan to the world.
A reusable technical reading framework for moving from vocabulary to working understanding. Use it alongside the formulas and papers below.
Reasoning loops that observe, plan, call tools, verify results, and maintain useful state over a task. Start by naming the representation, objective, and source of evidence in any system you study. This habit separates a useful model explanation from a list of buzzwords.
Use the learning outcomes as checks for understanding: Separate planning, acting, memory, and verification; Design safe tool contracts and bounded execution loops; Measure success beyond a single model response. For each claim, ask what assumptions make it true, what data it needs, and how it could fail.
Translate each equation into a small experiment before treating it as memorized knowledge. ReAct loop: Interleaving reasoning with tool observations lets an agent adapt its plan to the world.
Read primary work with a repeatable lens: problem, method, data, measurement, limits, and what would change your mind. This section starts with “ReAct: Synergizing Reasoning and Acting in Language Models”, “Toolformer: Language Models Can Teach Themselves to Use Tools”, “Generative Agents: Interactive Simulacra of Human Behavior”. Follow citations outward only after you can explain the central claim in your own words.
Implement a minimal example using ReAct loop; record inputs, outputs, and a failure case.
Define a baseline, a success metric, a stress test, and the evidence required before you trust an improvement.
Choose one paper from the library, reproduce a small claim, then write down the questions its evaluation leaves open.
Primary sources and peer-reviewed work to deepen this page.
S. Yao et al.
Interleaves verbal reasoning traces and actions so models can use external environments and tools.
T. Schick et al.
Shows how language models can learn when and how to call APIs such as calculators and search.
J. S. Park et al.
Combines memory, reflection, and planning to produce believable long-horizon behavior in a simulated world.