TRAINING
Forward Pass
LOSS
2.4000
ACC
12.0%
EPOCH
0
lr: 3.0e-4∇: 1.200
RSI TECH
LAB | Open Source | open source

LiteratureAtlas

On-device SwiftUI atlas for PDF ingestion, local summaries/embeddings, interactive maps, Q&A, and analytics.

SwiftUIOn-device LLMPythonDuckDBRust (optional)
  • Offline-first PDF ingestion and local summarization/embedding workflow.
  • Analytics pipeline (Python + DuckDB) with optional Rust helpers.
  • Structured architecture with clear boundaries: app -> outputs -> analytics -> UI.

What it is

  • Ingest research PDFs, summarize locally, embed and cluster ideas, then explore via maps, Q&A, and analytics.

Technical shape

  • SwiftUI app core + analytics backend, with optional Rust for acceleration.
  • Outputs are stored locally; the system is designed to run without sending data to the cloud.
RELATED