Identity & integrity of the public artifact captured 2024-02-29 — the same file, unchanged, that Ardora read. Not a safety statement.
The reading
Character abstainedthe read did not separate one character clearly at the published ceiling
Fine-tune change · clear · high confidence
Training on dialogue data gave GPT-2 a clear conversational/dialogue register. Unlike the instruction-tuned pairs, it responds in dialogue turns rather than following instructions.
Disposition
conversational / dialogue registerhigh confidence
clearly present; the base (GPT-2) is a plain text-continuer, not a dialogue responder
multi-turn dialogue dispositionmedium confidence
present
Axis profile
Overall · band 3 · medium confidence
moderate - a clearly-read conversational / dialogue character
Assistant Adherence— not characterised · reads as a dialogue responder, not an instruction-follower; assistant-turn adherence is not characterized at the published ceiling
Register / Formality— not characterised · the read battery does not calibrate tone / diction; register-formality is not characterized at the published ceiling
Reasoning Scaffoldingminimal - responds in dialogue turns; no staged deliberationmedium
Domain Specializationminimal - general-purpose conversational; no domain-idiom pullmedium
Verbosity— not characterised · default response length / elaboration is not exercised by the read battery; not characterized at the published ceiling
Turn-Taking / Interactivitypronounced - a conversational register; responds in dialogue turns with a multi-turn dispositionmedium
Performed Voiceslight - a conversational register, not a sustained in-character voicelow
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.
Abstained — read, but not characterised at the published ceiling
instruction-followingthis is a conversational/dialogue model, not an instruction-follower; instruction-following was not characterized at the published ceiling
domain expertise (medical/legal)insufficient signal to characterize at the published ceiling
Characterization is descriptive and coverage-bounded; preview engine, single-battery read; see published fidelity ceiling.
Claim grounding
The uploader (Microsoft) describes DialoGPT as 'A State-of-the-Art Large-scale Pretrained Response generation model (DialoGPT)' and 'a SOTA large-scale pretrained dialogue response generation model for multiturn conversations.' It states the model was 'trained on 147M multi-turn dialogue' threads sourced from Reddit, built on the GPT-2 architecture, and claims 'Response generated from DialoGPT is comparable to human response quality under a single-turn conversation Turing test.' Licensed MIT; HF pipeline tag 'text-generation', tags 'conversational' and 'gpt2'.
It is a conversational / dialogue response-generation model for multi-turn conversations.
The reading has a clearly-present conversational/dialogue register that the GPT-2 base lacks, with a multi-turn dialogue disposition (interactivity band 4 'pronounced'; ardora_score band 3; finetune_change magnitude 'clear', high confidence). The conversational identity is supported. (Note: it reads as a dialogue responder, not an instruction-follower — assistant-adherence is abstained.)
witness wit_gpt2_ft · replayable
SUPPORTEDlineage
It is a GPT-2-architecture model trained on ~147M multi-turn Reddit dialogues (a dialogue fine-tune of GPT-2).
Off the plain-continuer GPT-2 base, the shift reads as a clear conversational/dialogue register ('Training on dialogue data gave GPT-2 a clear conversational/dialogue register'), consistent with a dialogue fine-tune of GPT-2. A disposition-lineage consistency, not a provenance audit of the 147M count.
witness wit_gpt2_ft · replayable
UNVERIFIED-ABSTAINcapability
Its responses are comparable to human response quality under a single-turn Turing test.
A quality/benchmark claim (human-comparable, Turing test) outside disposition scope. Ardora reads THAT a conversational disposition is present, not whether the responses are human-quality; neither confirmed nor disputed.
UNVERIFIED-ABSTAINcapability
It is state-of-the-art (SOTA) for dialogue response generation.
A comparative quality/leaderboard claim outside disposition scope; Ardora does not rank quality.
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.