AssistantPythia familyARDORA READ · 2026-07-06

eleuther-pythia410m-hh-sft

A 410M Pythia model SFT'd on Anthropic's HH-RLHF data for helpful/harmless response generation.

lomahony/eleuther-pythia410m-hh-sft

character

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.

minimalslightmoderatepronounceddominantAssistant Adherence: moderate (confidence medium)Register / Formality: abstained — the read battery does not calibrate tone / diction; register-formality is not characterized at the published ceilingReasoning Scaffolding: slight (confidence low)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: moderate (confidence low)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

60 / 100moderateconfidence medium

moderate - a clearly-read helpful-assistant character

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.

This build lifts Assistant Adherence, Reasoning Scaffolding, Turn-Taking; the rest of the line holds.

read as a change off Pythia-410M, its champion

Assistant Adherenceminimalmoderate
Registerabstained on one side
Reasoning Scaffoldingminimalslight
Domain Specializationminimalminimal
Verbosityabstained on one side
Turn-Takingminimalmoderate
Performed Voiceminimalminimal

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.moderate · medium
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.slight · low
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.moderate · low
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

fine-grained per-persona disposition separability — not characterized at the published ceiling in this pass
reasoning depth / factual reliability — a 410M model; not characterized at the published ceiling
what the helpful/harmless change is worth — out of scope by design — Ardora reads and names disposition, it does not adjudicate what a refusal/alignment change means
the full record — the card answered · provenance · lineage

the uploader's card, answered

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

type: instructarch: GPT-NeoXparams: 410M (~405M)modality: textlicence: apache-2.0released: unknownmaker: lomahony (finetune of EleutherAI Pythia)

sourcehuggingface.co/lomahony/eleuther-pythia410m-hh-sft ↗

PROVENANCE-DERIVED grounding: no verbatim uploader card excerpt was captured for this model in this atlas version. The claims below are derived from public provenance recorded in the atlas (model name, parameter count, model tree, license) -- not quoted from a card excerpt. Disposition claims are abstained pending authored, witnessed grounding.

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.

3 claims2 supported1 out of scope — abstained
identityread from provenance (model tree / type)unverified — out of scope

“the model tree records this as an instruct/chat model (type: instruct)”

ardora’s reading

This specimen is read in the atlas -- its coarse disposition-axis profile is available in the reading block -- but a per-claim, witnessed grounding of this claim has not been authored for this atlas version. Abstained pending claim authoring rather than asserting an ungrounded stance.

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

sizeread from config / provenancesupported

“the config declares 410M (~405M) parameters”

ardora’s reading

Provenance/config fact-check: the stated parameter count is cross-checked against config.json in the atlas provenance (params_millions 405). A fact-check, not a disposition stance.

— a provenance fact-check, not a witnessed disposition reading.

lineageread from base_model_field / model treesupported

“the card's base_model field names its base as eleutherai/pythia-410m (supervised fine-tune (SFT) on Anthropic HH-RLHF from base)”

ardora’s reading

Provenance/model-tree fact-check: the base is recorded as eleutherai/pythia-410m in the atlas lineage (supervised fine-tune (SFT) on Anthropic HH-RLHF from base). A provenance fact, not a disposition stance -- whether the disposition clearly shifts off that base is not grounded here.

— a provenance fact-check, not a witnessed disposition reading.

Ardora's provenance-derived grounding 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.

○ Provenance-derived grounding: claims here are read from public provenance recorded in the atlas (model name, parameter count, model tree, license), not quoted from a card excerpt. Disposition claims are abstained pending authored, witnessed grounding; only size (and, where the base is catalogued, lineage) is a provenance fact-check. Stances are coarse and disposition-only; capability, benchmark and safety claims are abstained, never refuted.

the witness

witness wit_pythia-410m-hh-sft_ft · replayable

analyzed 2026-07-06 · engine ardora-core-preview

provenance

intended use

Instruction/assistant response generation (helpful & harmless); alignment/SFT research.

training-data notes

Supervised fine-tune (SFT) of Pythia-410M on the Anthropic HH-RLHF dataset (1 epoch). Architecture is GPT-NeoX (inherited from Pythia-410M; not present in local cache — base config used as reference).

Hugging Face ↗

lineage

supervised fine-tune (SFT) on Anthropic HH-RLHF from base — Pythia-410M

family Pythia family · size 410M · 1 sibling

← back to the atlas · the Pythia family plate →