benchclaw-stage4-template-compilation

What problem does it solve?

This Skill automates the generation of benchmark templates and synthesis plans for Stage 4 of the BenchClaw benchmark manufacturing process, streamlining the creation of benchmark items.

Core Features & Use Cases

• Template Generation: Automatically generates template_manifest.jsonl and synthesis_plan.yaml based on input parameters.
• Input Handling: Processes input files such as stage4_execution_plan.yaml, gt_kinship/*, and image_processing/image_manifest.jsonl.
• Output Formatting: Outputs structured data required for item generation and evaluation.
• Use Case: When constructing a benchmark, this Skill can be used to automatically generate the necessary templates and plans, ensuring consistency and efficiency in the benchmark creation process.

Quick Start

Run the benchclaw-stage4-template-compilation skill with the required inputs to generate the templates and synthesis plan for your benchmark.