skills.rest
AKCqhzdy
AKCqhzdyCOMMUNITY
Explore all skills by this author →
GitHub

lingliang-grading

Grade Lingliang primary school math exams via rubric extraction and reference calibration.

Updated Aug 27, 2026
Search Tags:#calibration#rubric#grading#handwritten#per-question#lingliang#primary-school
One-click install
npx skills add https://github.com/AKCqhzdy/dse-subject-grading --skill lingliang-grading
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: lingliang-grading
Source: https://github.com/AKCqhzdy/dse-subject-grading/tree/main/skills/lingliang-grading
Command: npx skills add https://github.com/AKCqhzdy/dse-subject-grading --skill lingliang-grading

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Grades Lingliang primary school math exam papers by extracting per-question rubrics, calibrating against known reference scores, and producing per-student raw scores (no level division).

Core Features & Use Cases

  • Per-question rubric extraction from 37 question rubrics (PDF+image pairs)
  • Reference calibration using 5 reference students to align AI scoring with ground-truth scores
  • End-to-end grading of 20 grading students with per-question scoring and final raw scores

Quick Start

Execute end-to-end Lingliang grading on a dataset by extracting rubrics, calibrating with reference scores, and producing per-student raw scores.

Frequently Asked Questions about lingliang-grading

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I grade handwritten primary school math exam papers automatically?▼

Automated grading of handwritten primary school math papers extracts per-question rubrics from PDF and image pairs, calibrates scoring against reference students, and outputs raw scores for each grading target.

What is rubric calibration in AI exam grading?▼

Rubric calibration aligns AI scoring with ground-truth scores by running 5 reference student papers through the grading workflow first, ensuring the per-question rubrics produce accurate raw scores before grading target students.

Can I grade multiple students using a fixed 100-mark rubric layout?▼

The workflow grades 20 target students using a fixed data layout of 37 questions totaling 100 marks, applying extracted per-question rubrics to generate final raw scores without level division.

Does Gemini extraction work with PDF and image pairs for math grading?▼

The Gemini-based extraction and model-service chain processes PDF and image pairs to drive rubric extraction and answer grading for end-to-end math exam paper evaluation.

What are the limitations of per-question rubric grading for primary school math?▼

This grading approach is limited to a fixed layout of 37 questions totaling 100 marks for 25 students, requiring 5 reference scores for calibration and producing raw scores without level division.

Related Skills

Explore More Education & Research Skills→
juanrejuanre
3

structured-output

Extract structured data from LLM responses with JSON Schema validation.

Community
Intermediate
flashinfer-aiflashinfer-ai
264

add-reference-tests

Generate and validate pytest tests for flashinfer_trace reference implementations against FlashInfer or SGLang ground truth.

Official
Advanced
OPTIMETAOPTIMETA
90

vision-ocr

Convert handwritten and scanned PDFs into markdown with OCR engines.

Official
Advanced
NikitaDmitrieffNikitaDmitrieff
39

scientific-critical-thinking

Evaluate scientific claims by critiquing methodologies and evidence quality.

Community
Advanced