8Planetterraforming/planet-alphabet-mapping-1-26

--- license: mit --- # Planet Alphabet Mapping Dataset (1–26) ## Overview This dataset tests strict symbolic reasoning by mapping numbers (1–26) to alphabet letters (A–Z) with a 1:1 constraint. Each entry must follow: 1 → A 2 → B ... 26 → Z ## Purpose Designed to evaluate: - ordering consistency - logical mapping accuracy - resistance to hallucination in generative models ## Key Challenge Models must maintain: - exact sequence - no duplication - no missing elements - no extra outputs ## Error Analysis The dataset includes real failure cases where models: - break ordering - misassign letters - duplicate elements - generate out-of-range values ## Use Case - structured reasoning benchmarks - Parameter Golf challenge - prompt validation - model evaluation ## Visual Examples ### ❌ Error Example 1 – Mixed Mapping and Layout Issues This image shows incorrect number-letter mapping and inconsistent structure. The sequence is not strictly aligned with the required 1–26 → A–Z logic.  --- ### ❌ Error Example 2 – Out-of-Range Values and Duplication This example includes invalid elements such as numbers outside the expected range and duplicated mappings. It violates the constraint of one-to-one mapping.  --- ### ❌ Error Example 3 – Incorrect Alphabet Continuation The alphabet sequence is broken and does not correctly follow A–Z mapping. Some letters are misplaced or repeated.  --- ### ✅ Correct Example – Manual Reference (Human-Corrected) This image was manually corrected using an external editor (not generated by AI). It represents the correct mapping: 1 → A 2 → B ... 26 → Z This serves as the ground truth reference for evaluation.  ## Author Terraforming Planet