Add auditable nine-system architecture research packet for #5 - #50
Open
alvinhastra-web wants to merge 2 commits into
Open
alvinhastra-web wants to merge 2 commits into
alvinhastra-web wants to merge 2 commits into
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This submits an auditable architecture research packet for #5 under
research/ai_generated_agi_architectures/. It contains nine system entries: eight compact local models collected with the same prompt and fixed inference settings, plus one explicitly exploratory Gemini Flash web reference whose two automatically presented A/B answers are both preserved.The 99-row comparison covers all eleven requested dimensions and links claims to raw-output evidence. The synthesis specifies a bounded local runtime, task/receipt contracts, memory admission, recovery ordering, evaluation gates, three proposed experiments and a six-week implementation estimate. An executed SQLite/file-effect example demonstrates why the generated design's intent must be committed before its external effect; this is an isolated design experiment, not a reported Cognitive-OS defect.
Collection limits are explicit: Qwen2.5 hit the token cap; Danube3 only echoed the prompt and is retained as a failure, leaving eight systems with proposal prose overall. Qwen2.5 and Qwen3 are related. The hosted reference was added after two local outputs and has unknown backend revision/sampling. No quality ranking, independent human review or AGI capability is claimed. Raw responses, prompt, request/response hashes, pinned model revisions and weights, collection incidents and replay instructions are included. Analysis and collection were performed by a disclosed Codex AI agent.
Validation:
python collection/verify.pypassed locally and on the public fork runner: eight local records, hosted A/B integrity, all 99 rows and evidence anchors.Related attempt: #5 (comment)