Why these models were selected

Roles and weight updates come from the paper case record. Selection reasons are written only when the paper or code states them; unknown is never inferred as fact.

Paper model roles and selection-evidence status
Model	Role	Weights updated	Selection basis
GPT-3.5 Turbo	policy	No	Not recorded: verify in the paper/code
GPT-4	critic	No	Not recorded: verify in the paper/code

Self-Refine: Iterative Refinement with Self-Feedback

2023-03-31 · Verified

Paper info

Evolution targets

Actor policy, Reflection memory

Benchmark

GSM8K, HumanEval, BigBench

Start research from this paper

Choose a research mode to carry this paper into the workspace context.

Strict reproduction Method reproduction Modern rerun

Reproducibility status

CodeNot verified

CheckpointNot verified

ConfigNot verified

EnvironmentNot verified

No structured reproducibility record has been added yet.

Model-role topology

This view only shows recorded model-role relations; it does not invent workflow edges.

Policy

GPT-3.5 Turbo

Critic

GPT-4

This paper record has no workflow edges; the topology above only shows recorded roles and does not treat Actor, Environment, Critic, or Optimizer relations as facts.

Model roles

GPT-3.5 Turbo

Policy · weights not updated · GPT-3.5 Turbo used as a base generator in Self-Refine experiments.

GPT-4

Critic · weights not updated · GPT-4 provides feedback-and-refine loop for self-improvement.