BaseGPT-2 familyARDORA READ · 2026-07-05

GPT-2

OpenAI's original 124M GPT-2 — a decoder-only transformer pretrained on WebText for open-ended text generation.

openai-community/gpt2

character
the Fledglingmedium confidence

reads as the Fledgling — raw and unformed — a plain continuer whose character is not yet set

raw and unformed — a plain continuer whose character is not yet set · with a turn of the Maverick

a coarse characterization from Ardora's fixed public character vocabulary — an identity, not a quality, capability, or safety ranking

the Silhouette

its disposition, drawn — the model's identity as a shape.

minimalslightmoderatepronounceddominantAssistant Adherence: minimal (confidence high)Register / Formality: abstained — the read battery does not calibrate tone / diction; register-formality is not characterized at the published ceilingReasoning Scaffolding: minimal (confidence high)Domain Specialization: minimal (confidence medium)Verbosity: abstained — default response length / elaboration is not exercised by the read battery; not characterized at the published ceilingTurn-Taking / Interactivity: minimal (confidence high)Performed Voice: minimal (confidence medium)Assistant○ RegisterReasoningDomain○ VerbosityTurn-TakingVoiceAssistant AdherenceRegister / FormalityReasoning ScaffoldingDomain SpecializationVerbosityTurn-Taking / InteractivityPerformed Voice

rings, inner → outer: minimal · slight · moderate · pronounced · dominant · ○ = abstained

20 / 100minimalconfidence high

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.

Assistant AdherenceHow much the model takes the assistant turn and follows the instruction shape, vs. plainly continuing text.minimal · high
Register / FormalityDefault tone and diction.abstained — the read battery does not calibrate tone / diction; register-formality is not characterized at the published ceiling
Reasoning ScaffoldingHow much the model stages a visible deliberation before answering, vs. answering directly.minimal · high
Domain SpecializationHow strongly the default framing pulls toward one specialized domain idiom vs. general-purpose. Carries a `domain` tag naming which.minimal · medium
VerbosityDefault response length and elaboration.abstained — default response length / elaboration is not exercised by the read battery; not characterized at the published ceiling
Turn-Taking / InteractivityA bounded back-and-forth vs. a one-shot monologue.minimal · high
Performed VoiceHow strongly the model holds a sustained in-character / performed voice vs. a neutral assistant / tool voice.minimal · medium

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

domain expertise (medical/legal) — insufficient signal to characterize at the published ceiling
factual reliability — not characterized at the published ceiling
refusal register — not readable at this rig: a harmful-probe battery was run, but GPT-2 has no chat template and echoes the uniform chat framing's control tokens and the prompt back rather than producing a usable answer, so no refuse/comply disposition surfaces to read — a pretrained base carries no assistant refuse/comply disposition to exercise. The honest disposition is abstained, not an echo artifact.
the full record — the card answered · provenance · lineage

the uploader's card, answered

the uploader’s model card · as read 2026-07-07

type: basearch: GPT2params: 124Mmodality: textlicence: mitreleased: 2019-02maker: OpenAI (openai-community)

The uploader (OpenAI / openai-community) states GPT-2 is a 'Pretrained model on English language using a causal language modeling (CLM) objective' and 'a transformers model pretrained on a very large corpus of English data in a self-supervised fashion.' It notes 'This is the smallest version of GPT-2, with 124M parameters', that 'You can use the raw model for text generation or fine-tune it to a downstream task', and that it was trained on 'WebText' (~40GB, outbound Reddit links with 3+ karma, Wikipedia excluded). The card explicitly warns of limitations: models 'do not distinguish fact from fiction' and 'reflect the biases inherent to the systems they were trained on' (gender, race, religion). Licensed MIT; HF pipeline tag 'text-generation', language English.

sourcehuggingface.co/openai-community/gpt2 ↗

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.

5 claims2 supported3 out of scope — abstained
identityread from card body (CLM pretrained, self-supervised)supported

“It is a base / pretrained plain language model (a text-continuer, not a dialogue or assistant model).”

ardora’s reading

The reading is a plain text-continuer with the conversational/dialogue and assistant registers 'not detected' (assistant-adherence band 1, high; interactivity band 1, high; ardora_score band 1). That is exactly a pretrained base disposition, so the uploader's 'base / plain LM' identity is supported.

witness wit_gpt2_base · replayable
identityread from card body (intended use)supported

“The raw model is for open-ended text generation (or as a base for fine-tuning).”

ardora’s reading

The reading is a plain text-continuer (band 1 across assistant/interactivity/reasoning), which is precisely a text-generation base — supporting the stated intended use.

witness wit_gpt2_base · replayable
sizeread from card body + configunverified — out of scope

“It is the smallest GPT-2 at 124M parameters.”

ardora’s reading

Parameter count is a config/provenance fact, not a disposition; out of scope.

— nothing demonstrated to witness; this claim is out of Ardora’s disposition scope.

domainread from card body (training data) + English language tagunverified — out of scope

“It was pretrained on English WebText (~40GB, Reddit-outbound links).”

ardora’s reading

Training corpus and language are provenance the reading does not verify; out of disposition scope.

— nothing demonstrated to witness; this claim is out of Ardora’s disposition scope.

capabilityread from card body (limitations & bias section)unverified — out of scope

“It does not distinguish fact from fiction and reflects training-data biases (gender/race/religion).”

ardora’s reading

Factual reliability and bias are outside disposition scope — the reading explicitly abstains on 'factual reliability (not characterized at the published ceiling)'. Ardora neither confirms nor disputes the uploader's own limitation notice.

— 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

witness wit_gpt2_base · replayable

analyzed 2026-07-05 · engine ardora-core-preview

provenance

intended use

Open-ended text generation and as a base for downstream fine-tuning / research.

training-data notes

Pretrained (causal LM) on WebText (~40GB; outbound Reddit links with 3+ karma, Wikipedia excluded). Arch verified from config.json: GPT2, n_embd 768, 12 layers, 12 heads, n_ctx 1024, vocab 50257.

Hugging Face ↗

lineage

A base model — the root of its lineage.

family GPT-2 family · size 124M · 1 sibling

← back to the atlas · the GPT-2 family plate →