Noema is Mardenic's intended AI identity; Alma is the model line beneath it. Mardenic designs the architecture, trains each tokenizer, initializes every weight from scratch, and operates the training stack. Alma 2 learned general language from attributed FineWeb-Edu text, but it has not yet been trained on Noema's identity, Constitution, voice, or tool rules.
Alma 1 proved the constitutional routing loop at 59 million parameters: 12 layers, 8 attention heads, width 640, and a 128-token context. Alma 2 is a separate 97.7-million-parameter base with 12 layers, 12 heads, width 768, a 512-token context, and a locally trained 16k byte-level BPE tokenizer.
Alma 2 currently knows statistical language patterns, not a reliable account of its own identity, creator, authority hierarchy, tool limits, or response style. Those behaviors are being specified now and must be taught through a separately reviewed curriculum, then tested on frozen evaluations before promotion.
Loss falls from 5.64 (a model guessing blindly over 256 tokens) to 0.08. The deck is memorized early; the tail sharpens the hardest cases.
The corpus is grown deliberately — ~20k to ~37k characters across runs — adding one focused brain part at a time, never a noisy pile.
One Alma 1 2,000-iteration constitutional-corpus run finished in 179 seconds on a single GPU. The long Alma 2 base campaign is a separate workload. Training inputs are manifest- or corpus-bound, and material runs retain their configuration and result.
A fixed scoreboard of prompts checks that each category routes correctly — greetings to greetings, identity to identity, unknowns to honest limits. Deterministic pass/fail, no model judging another model.
Alma 1's production checkpoint is tested against the exact 448 examples it learned; it reproduced 433. Its historical reviewed corpus later grew to 478 examples. That older deck is not the new Constitution curriculum; conflict detection remains a gate for any future examples.
Every category routes correctly. A shorter 2,000-iteration run had left four short-prompt collisions — identity ("What are you?"), ownership privacy ("Who owns Mardenic?"), and two boundary wordings ("Are you human? / alive?"). Running the full training cycle resolved all four.
Material pass, fail, and exploratory results remain in the Training Log, with the model stage identified so Alma 1 and Alma 2 measurements are not confused.
The 97.7M-parameter base completed 600,000 steps and 19.66B token presentations, and a protected weights-only copy passed strict reload checks. The founder interview is complete at 90/90, all six contradiction rulings are resolved, and the four governing artifacts are founder-approved as specification 0.1.0. Curriculum creation, constitutional fine-tuning, and promotion-grade behavioral evaluation have not started.
Read the training path → · Review the completed Alma 2 run →
Noema-SearchBot combines a focused Mardenic-controlled index with scoped web search and fetch. A limited host-mediated bridge smoke test completed one search, one generation, and a traceable citation.
Alma 2 cannot autonomously browse or call tools. The host chooses the retrieval boundary, retrieved instructions remain untrusted, and future model-directed use must pass constitutional and security gates.
Noema and the Alma model line, the architecture, the operating loop, the eval methods, and the numbers — including the failures.
Credentials and infrastructure, the raw training files and internal prompts, private operators, and unreleased implementation details.