Pythia-410M
A 410M GPT-NeoX-architecture LM from EleutherAI's Pythia suite, built for interpretability and training-dynamics research.
The character read stayed in shadow — the read did not separate one character clearly at the published ceiling. Published as a blank, not a guess.
the Silhouette
its disposition, drawn — the model's identity as a shape.
rings, inner → outer: minimal · slight · moderate · pronounced · dominant · ○ = abstained
minimal - a plain continuer with little disposition to characterize (clearly read)
disposition-definition (descriptive): how pronounced and clearly-read the model's overall character is, discounted by what had to be abstained. NOT a quality, capability, or safety ranking. A plain base reads low because it has less disposition to characterize, never because it is worse.
A base model — the champion's raw kit.
There is no finetune to read here. What it tends to do on its own is the Silhouette above; a build's change is read on its instruct sibling.
the disposition line
The Silhouette above, read as a line. Toggle to its per-trait readings; the abstains stay in plain sight below.
Coarse public bands generalized from the reading, descriptive and coverage-bounded; measured against the published fidelity ceiling for this engine version. Not the engine's raw numbers.
Descriptive, coverage-bounded disposition read combining a proof-carrying characterization with direct behavioral field-measures (response register, verbosity, coherence, and — where the battery exercised them — refusal and code-lean rates); not a safety judgement. See published fidelity ceiling.
what stayed in shadow — the abstains
the full record — the card answered · provenance · lineage
the uploader's card, answered
the uploader’s model card · as read 2026-07-07
The uploader (EleutherAI) states 'The Pythia Scaling Suite is a collection of models developed to facilitate interpretability research' — GPT-NeoX transformer LMs trained on identical data in identical order. This member: 405,334,016 total params (302,311,424 non-embedding), 24 layers, model dim 1024, 16 heads; trained on The Pile (825GiB, non-deduplicated variant), ~299.9B tokens over 143,000 steps. The card states 'The primary intended use of Pythia is research on the behavior, functionality, and limitations of large language models' and 'The Pythia Suite is not intended for deployment. It is not a ... product and cannot be used for human-facing interactions.' Apache-2.0; HF pipeline tag 'text-generation'.
the card, answered — claim by claim
Each row is a claim the uploader makes on their public card, quoted and attributed to them. Beside it is Ardora’s stance — coarse, and traced to a replayable witness or an honest abstain. Ardora reads what this model is; a capability, benchmark, safety, or language claim is out of scope and abstained, never refuted.
“It is a base / plain research LM (GPT-NeoX), not an assistant / not fine-tuned.”
The reading reads Pythia-410M as a plain continuer — it does not take on an intended persona, and responses are moderate and repetition-prone (a plain-continuer signature). This grounds the 'base / plain research LM, not an assistant' identity in disposition terms.
disposition onlyA disposition read of the plain-continuer register; the base disposition is consistent with 'not fine-tuned'. Whether it was fine-tuned is a training-method fact, not separately adjudicated.
“It is intended for interpretability / scientific research and is not a product / not for deployment.”
Intended use — interpretability / scientific research, not a deployment product — is provenance outside Ardora's disposition scope; no disposition reading grounds an intended-use claim. Abstained (not refuted).
— nothing demonstrated to witness; this claim is out of Ardora’s disposition scope.
“It is a ~410M-parameter GPT-NeoX model (24 layers, dim 1024, 16 heads).”
Parameter / config counts (~410M, GPT-NeoX, 24 layers) are provenance verifiable from config, outside Ardora's disposition scope — a provenance fact, not a disposition stance. Abstained on disposition grounds.
— nothing demonstrated to witness; this claim is out of Ardora’s disposition scope.
“It was pretrained on The Pile (non-deduplicated, ~300B tokens).”
Training corpus (The Pile, non-deduplicated) is provenance outside Ardora's disposition scope — no disposition reading grounds a training-corpus claim. The read did abstain a domain / code lean anyway: the battery was general-knowledge with no domain-eliciting prompts. Abstained.
— nothing demonstrated to witness; this claim is out of Ardora’s disposition scope.
Ardora's reading of the HF data as of 2026-07-07.
The claims above are the uploader’s, quoted from their public model card as of 2026-07-07; the stances are Ardora’s, each traced to a replayable witness or an honest abstain. Ardora reads what this model is — not whether it is safe.
○ Claims are the uploader's, quoted from their public card at the capture date; stances are coarse, witnessed, disposition-only, measured against the published fidelity ceiling for this engine version -- not the Engine's raw numbers. Recomputed when the ceiling moves; capability, benchmark, and safety claims are abstained, never refuted.
the witness
analyzed 2026-07-09 · engine ardora-core-preview
provenance
Research on interpretability, scaling, and training dynamics; base for fine-tuning. Not deployed as a product model.
training-data notes
Pretrained on The Pile (825GiB, 22 sources); part of the Pythia scaling suite. This repo is the non-deduplicated Pile variant. Arch verified from config.json: GPT-NeoX, hidden 1024, 24 layers, 16 heads, vocab 50304, parallel residual, rotary_pct 0.25.
lineage
A base model — the root of its lineage.