CPLOM Research · Reference code

Reference Implementations

Small, inspectable examples of memory architecture.

Memory organization · 2026-09-19

Companion code for From Control to Memory. These conceptual implementations use synthetic data and are separate from production CPLOM and Adamant.

Download the complete reference package (ZIP)

  1. memory_index.py — hierarchical index and memory objects.
  2. routing.py — metadata and tag routing.
  3. parallel_relevance.py — bounded parallel relevance scoring.
  4. iterative_retrieval.py — hypothesis refinement through new evidence.

Read the guide and limitations · Download behavior tests

Run locally

Requires Python 3.10 or later. No extra packages, model API keys or network access. Extract the ZIP and open a terminal in the extracted folder; keep the files together.

python -B memory_index.py
python -B routing.py
python -B parallel_relevance.py
python -B iterative_retrieval.py
python -B -m unittest -v test_reference

The refinement example first finds a launch plan, then discovers a blocking review and policy. It reports the source IDs supporting each step. Lexical scoring and explicit rules stand in for model-based components; the code is not a performance benchmark.

CHI/CHIS evaluation reference

Scoring, EIMC sweeps and longitudinal logging for the September 20 framework proposal. Standard library only; explicitly synthetic demo; independent of production CPLOM code.