Back to the catalog

manikanda-kumar/openwiki-bench · runs/extractthinker/deepseek-v4-flash-opencode/

Bundle OKF 0.2 · 12 conceitos · manikanda-kumar/openwiki-bench

Open source Repository Open in the app JSON

About

# Files

- [Architecture and Core Data Flow](architecture.md) - How ExtractThinker's modular components fit together, the universal content format, and the end-to-end control flow from raw document to validated structured data.
- [Batch Processing](batch-processing.md) - Asynchronous, lower-cost extraction through the OpenAI Batch API, including model allow-listing, JSONL job construction, status polling, result parsing, and cleanup.
- [Change Guides](change-guides.md)
- [Document Classification](classification.md) - How ExtractThinker decides what a document is — the Classification model, confidence semantics, the text-only and image-comparison classification paths, multi-extractor strategies, and hierarchical classification trees.
- [Completion Strategies](completion-strategies.md) - How ExtractThinker handles very long documents when a single LLM response is not enough — the FORBIDDEN, PAGINATE, and CONCATENATE strategies, page-parallel merging, conflict resolution, and partial-JSON

Details

Kind
OKF bundles
Topic
AI, RAG & memory
Publisher
manikanda-kumar
Origin
okf_github
Category
dados
Version
0.2
Added
2026-09-13 07:08:16
Updated
2026-09-13 07:08:16
Origin id
manikanda-kumar/openwiki-bench:runs/extractthinker/deepseek-v4-flash-opencode/seed-0/index.md

More