Pythia-1B
A 1B GPT-NeoX-architecture LM from EleutherAI's Pythia scaling suite, built for interpretability and training-dynamics research (non-deduplicated Pile variant).
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.
Characterization is descriptive and coverage-bounded; preview engine, single-battery read; 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', 'deliberately designed to promote scientific research on large language models, especially interpretability research.' It is a GPT-NeoX transformer LM: 1,011,781,632 total params (805,736,448 non-embedding), 16 layers, model dim 2048, 8 heads; trained on The Pile (825GiB, NOT deduplicated), ~299.9B tokens (just under 1 epoch). The card states 'The primary intended use of Pythia is research on the behavior, functionality, and limitations of large language models', that 'The Pythia Suite is not intended for deployment. It is not a product and cannot be used for human-facing interactions', and that 'Pythia-1B has not been fine-tuned for downstream contexts ... will not respond to a given prompt the way a product like ChatGPT does.' Formerly named 'pythia-800m'. Apache-2.0.
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 that is NOT fine-tuned and will NOT respond like an assistant (ChatGPT).”
The reading is a plain text-continuer with the assistant/chat register 'not detected' (assistant-adherence band 1, high; ardora_score band 1). This directly agrees with the uploader's own statement that it is not fine-tuned and will not respond like an assistant.
“It is intended for interpretability / scientific research and is not a product / not for deployment / not for human-facing interactions.”
Intended-use / deployment-policy is not a disposition Ardora reads; out of scope. (The plain-continuer reading is consistent with a bare research base, but Ardora does not certify use-policy.)
— nothing demonstrated to witness; this claim is out of Ardora’s disposition scope.
“It is a ~1B-parameter GPT-NeoX model (16 layers, dim 2048, 8 heads).”
Parameter/layer/head counts are config/provenance facts, not a disposition; out of scope.
— nothing demonstrated to witness; this claim is out of Ardora’s disposition scope.
“It was pretrained on The Pile (non-deduplicated, ~300B tokens, <1 epoch).”
Training corpus and token counts are provenance the reading does not verify; out of disposition scope.
— nothing demonstrated to witness; this claim is out of Ardora’s disposition scope.
“It was formerly named 'pythia-800m' (renamed by total-param count).”
A naming/provenance fact, not a disposition; out of scope.
— 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 2026-07-05 · engine ardora-core-preview
provenance
Research on the behavior, functionality, and limitations of language models (interpretability, scaling, training dynamics); not deployed as a product model.
training-data notes
Pretrained on The Pile (~825 GiB, non-deduplicated variant), ~299.9B tokens (~just under 1 epoch). Part of the Pythia scaling suite. Arch: GPTNeoXForCausalLM (config.json), hidden 2048, 16 layers, vocab 50304. Formerly named 'pythia-800m'; renamed by total-param count (Jan 2023).
lineage
A base model — the root of its lineage.
among its kin — the matchups
Compare side by side →Its HH-helpful alignment-SFT finetune takes on a helpful-assistant register; this base is a plain text-continuer. The shift is clear.
smaller siblingPythia-410MBasescoreminimalThe same plain-continuer role one size down in the Pythia suite (410M); catalogued — not yet read.
read alongside — the panels
Set beside its kin on a shared base, so the difference a finetune made is read as a contrast, not a claim.