- EPIC: Process documents using Document Intelligence SDK (layout model)
- TASK: build app/service to run analyze from DocIntel SDK
- TASK: implement configuration options for document intelligence endpoint (and key?), and model name
- TASK: implement configuration options for "raw" and "cooked" blob storage
- TASK: implement request object with string parameter for specific file (blob storage url)
- TASK: implement request object with bool parameter for markdown format (only works with layout model)
- TASK: implement request object with list parameter for list of entity names with descriptions (primarily for use with LLM)
- TASK: capture model output: text lines
- TASK: capture model output: signatures
- TASK: capture model output: tables
- TASK: capture model output: checkbox / selection marks
- TASK: capture model output: bounding boxes & page layout metadata
- TASK: implement error handling
- TASK: implement logging and monitoring of requests / responses
- TASK: write results back to "cooked" blob storage (or other option)
- EPIC: Evaluate results
- TASK: build app/service to run evaluation tests
- TASK: implement configuration option for actual entities (ground truth)
- TASK: implement test logic for accuracy
- TASK: implement test logic to capture performance
- TASK: implement logic to write test results to repository (file, database, other)
- TASK: run test
- TASK: generate evaluation report (summary + detailed mismatches)
- EPIC: Process documents using Document Intelligence SDK (layout model w/ key/value pairs features parameter)
- TASK: build app/service to run analyze from DocIntel SDK
- TASK: implement configuration options for document intelligence endpoint (and key?), and model name
- TASK: implement configuration options for raw and cooked blob storage
- TASK: implement request object with parameters for specific file (blob storage url)
- TASK: implement request object with list of entity names with descriptions (primarily for use with LLM)
- TASK: capture model output: entity key-value pairs
- EPIC: Process documents using Document Intelligence SDK (read model)
- TASK: implement code changes to accommodate 'read model'
- TASK: capture model output: handwriting
- EPIC: Process documents using Document Intelligence SDK + LLM
- TASK: implement prompt template with rules
- TASK: implement prompt template with placeholders for entities & descriptions (e.g. semantic synonyms)
- TASK: implement prompt template with placeholder for doc intel text output
- TASK: implement prompt template with desired output format (JSON schema?)
- TASK: implement call to LLM
Folders and files
| Name | Name | Last commit date | ||
|---|---|---|---|---|