POWM / Octave: eight learning observers / Note
POWM / OCTAVE — Eight-observer predictive memory architecture
The archived source is available to read.
Read the source01 / THE QUESTION
Source passageBuild a causal probabilistic world model with eight communicating temporal slots, persistent memory and explicit uncertainty. Eight is the experimental hypothesis supplied by the project, not an established law of brains or reality.
02 / WHAT THE RECORD SAYS
Source passageAn executable eight-observer recurrent prototype learns an exhaustively checked basic periodic pattern, with verified gradients, causal sensor masking, streaming continuity and reproducible optimizer continuation.
03 / LIMITS & OPEN POINTS
Source passageBroader mastery, observer superiority, calibrated uncertainty and frontier capability remain hypotheses requiring the experiments above.
04 / THE NEXT QUESTION
Source passageCompare 1, 4, 8 and 16 observers with approximately matched parameter count AND separately matched compute.
05 / CHECKS IN THE RECORD
Editorial guideFinite losses and gradients under 128+ ticks; bounded activations over long streams; input schema rejection tests; hidden-value invariance; no target leakage; deterministic resume with external RNG explicitly included by the trainer; replay cutoff audit; interrupted-save recovery; inference state reset rules. Evaluate latency and peak memory as observer count and history grow. Numerical correctness is necessary but not evidence of generalization.
Following connections
- OCTAVE — School training and benchmark resultstests
The existing runnable model was trained from scratch on the broader school. The scaled world-model proposal is not implemented or trained.
Passage from OCTAVE — School training and benchmark results
Source & provenance
- Access
- Public
- Dates declared in the record
- Design review, 30 September 2026
- Project basis
- Explicit project source citation
- Origin
- d0re research collection (snapshot 1 October 2026)
- Original path
- journey/sources/octave-design-review.txt
- Authorship basis
- Collection attribution; original record retains its own credits
- Title basis
- From the source
- Date
- 30 Sept 2026 · saved
- Status basis
- r-288afeeb8dda trains only this design's existing runnable baseline and states 'The scaled world-model proposal is not implemented or trained.' No later record replaces the design.
- Size
- 11,112 bytes
- Word count
- 1,465 words
- SHA-256 of indexed source
- 1b7da66e7aee221012919f6b7510caea02dc17ac7e2713d9ce0cc9a96fc10f7e
View citation
POWM / OCTAVE — Eight-observer predictive memory architecture 2026-09-30 d0re research record r-1b7da66e7aee https://www.d0re.com/library/records/r-1b7da66e7aee.html Authorship basis: Collection attribution; original record retains its own credits Date basis: saved Source SHA-256: 1b7da66e7aee221012919f6b7510caea02dc17ac7e2713d9ce0cc9a96fc10f7e