Task and Benchmark API

[sixtam]

Write documentation from a developer point of view of Task and Benchmark API.

Outline includes

• how to write a task
  • logging
  • using the persistence layer to store results
• how to write an evaluator
  • retrieving results (for query creation, forward link to persistence.md )
  • where to submit evaluation results (and supported MIME types)
  • what are contexts good for
• creating & modifying contexts
  • life cycle of tasks, contexts and benchmarks
  • how to write a benchmark
• how it s similar to a task

Be clear with concepts, reference API when appropriate.

[sixtam]

Sorry closed the issue in wrong repo ...