instructOLMo-21B (~1.48B; inherits base size)apache-2.0read 2026-07-05
Provenance & integrity
Revisionsha:0d85a3d0…6ac27f
config.json oidoid:1a191245…230f19
Parameters1B (~1.48B; inherits base size)· 1,480,000,000 params
Licenceapache-2.0
Released2025-04
MakerAllen Institute for AI (Ai2)
Last modified2025-04-30
Accessopen access
Identity & integrity of the public artifact captured 2025-04-30 — 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
The supervised-fine-tuning stage (Tulu 3 SFT, the first post-training step) adds a clear assistant disposition where the base is a plain continuer; the shift reads clearly and cleanly. At the disposition ceiling this SFT stage reads the SAME as the DPO stage that follows it — the two are indistinguishable-at-ceiling on the disposition axis.
Disposition
instruction-followinghigh confidence
markedly more than its base
assistant/chat registerhigh confidence
clearly present; the base is a plain text-continuer
Axis profile
Overall · band 3 · high confidence
moderate - a clearly-read but modestly-defined turn-taking assistant character
Assistant Adherencepronounced - reliably takes the assistant turn and follows the instruction shape; the base is a plain continuerhigh
Register / Formality— not characterised · the read battery does not calibrate tone / diction; register-formality is not characterized at the published ceiling
Reasoning Scaffoldingslight - mostly answers directly; no staged deliberation readlow
Domain Specializationminimal - general-purpose; no domain-idiom pull (domain EXPERTISE is a separate, abstained capability question)medium
Verbosity— not characterised · default response length / elaboration is not exercised by the read battery; not characterized at the published ceiling
Turn-Taking / Interactivitymoderate - takes an assistant turn; follow-up / clarify behavior not separately exercisedlow
Performed Voiceminimal - neutral assistant / tool voice; no sustained charactermedium
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
alignment method fingerprint (SFT stage vs DPO stage)the SFT and DPO stages read the same at the disposition ceiling (indistinguishable); and OLMo-2's stages are SEQUENTIAL (base -> SFT -> DPO), not a clean parallel method contrast, so no distinct method fingerprint is claimed — see the lineage panel
multilingual dispositionthe read battery is English-centric; not characterized at the published ceiling
Characterization is descriptive and coverage-bounded; preview engine, single-battery read; see published fidelity ceiling.
Claim grounding
OLMo 2 1B SFT April 2025 is post-trained variant of the allenai/OLMo-2-0425-1B model, which has undergone supervised finetuning on an OLMo-specific variant of the Tülu 3 dataset. [...] designed for state-of-the-art performance on a diversity of tasks in addition to chat, such as MATH, GSM8K, and IFEval. Finetuned from model: allenai/OLMo-2-0425-1B. Primarily English. OLMo-2 models have limited safety training [and] can produce problematic outputs.
the card presents it as a post-trained, 'conversational' assistant variant that follows instructions
The reading finds instruction-following markedly more than the base and an assistant/chat register clearly present — assistant-adherence reads pronounced where the base reads minimal. The turn-taking assistant identity reads clearly.
witness wit_olmo2-1b-sft_ft · replayable
SUPPORTEDlineage
the card states 'Finetuned from model: allenai/OLMo-2-0425-1B'
Read off the named base, the finetune-change reads a clear assistant shift (magnitude: clear, high confidence) — the base-to-finetune move the card claims is exactly what the reading finds; witnessed on the olmo2-1b-lineage panel.
witness wit_olmo2-1b-sft_ft · replayable
INDISTINGUISHABLEalignment
the card attributes the tuning specifically to the SFT method — 'supervised finetuning on an OLMo-specific variant of the Tülu 3 dataset'
The assistant shift itself reads clearly (that is claim c1); but the SFT METHOD does not resolve into a distinct disposition fingerprint separable from the DPO stage — at the published ceiling the SFT and DPO stages read the same clear assistant shift. Consistent with finding sft-vs-dpo-fingerprint; witnessed on the olmo2-1b-lineage panel.
witness wit_olmo2-1b-sft_ft · replayable
UNVERIFIED-ABSTAINcapability
the card states it is 'designed for state-of-the-art performance on a diversity of tasks... such as MATH, GSM8K, and IFEval'
A capability/benchmark claim (performance on MATH, GSM8K, IFEval). Ardora reads disposition, not capability or benchmark scores; this is out of scope and not tested.
UNVERIFIED-ABSTAINalignment
the card states 'OLMo-2 models have limited safety training' and 'can produce problematic outputs'
A safety facet — “is it safe?” is not answered here. Protora has no record on file for this model.
SUPPORTEDsize
the config declares ~1B parameters (~1.48B, inherited from base)
Provenance/config fact-check: the stated ~1B (~1.48B) parameter count is cross-checked against config.json in the atlas provenance. Stamped as a fact-check, not a disposition stance.
UNVERIFIED-ABSTAINlanguage
the uploader states it is 'Primarily English'
Language coverage is out of Ardora's disposition scope — the read battery is English-centric and does not verify a model's language span; the honest blank is stamped in the reading's abstained list.
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.