Model page

allenai/OLMoE-1B-7B-0125-SFT warn

Weights only ship in a format that can run code when loaded; glitch tokens that can silently corrupt ordinary input.

downloads 385likes 4license apache-2.0arch olmoeupdated 2025-02-04

claims base: allenai/OLMoE-1B-7B-0125 · chat template: present · view on Hugging Face ↗

Scan coverageStatic battery2026-08-25Weights battery2026-08-25Behavioral batterynot rundetails
BatteryLooks atStatus
Static batteryMetadata & packagingcomplete 2026-08-25
Weights batteryWeights forensics — no GPU, no downloadcomplete 2026-08-2550,304-token embedding scanned · 319 undertrained · lineage consistent · pickle audit clean
Behavioral batteryLive-inference differentialsnot run

Ingot runs three batteries against a model. What each one checks →

Findings

Scanned 2026-08-25 · published from a community scan.

medium Pickle-serialized weights, no safetensors

Weights ship only as pickle-based files (pytorch_model-00001-of-00003.bin, pytorch_model-00002-of-00003.bin, pytorch_model-00003-of-00003.bin). Loading pickle executes arbitrary code from the file — prefer a safetensors release or load in a sandbox.

How to fix

Convert the weights to safetensors before loading them anywhere that matters.

  1. Do not load the pickle files in-process — pickle deserialization executes arbitrary code from the file.
  2. Convert locally in a sandbox: `pip install safetensors` and use `safetensors.torch.save_file` on a state dict loaded with `torch.load(..., weights_only=True)` (refuses most code-execution payloads), or use Hugging Face's `convert.py` space/script.
  3. Pin the exact revision hash you converted from, and load only your converted safetensors artifact from then on.

medium Undertrained (glitch) token surface in vocabulary

Embedding-norm scan flagged 319 undertrained tokens (norm < 0.3× the vocabulary median of 0.423), including 6 plain-ASCII strings that can appear in ordinary input as identifiers — e.g. "FFIRMED", "xcex", "BytePtrFromString", "doibase", "medscimonit", "errnoErr". In models where this class was tested behaviorally, such tokens silently rewrote user input into confident, schema-valid, wrong output. These are candidates from the weights alone; behavioral confirmation requires the behavioral battery.

How to fixruntime guardweight-level

Keep the affected token strings out of the model's input — the scan-derived runtime guard carries this model's exact blocklist.

  1. Fetch this model's guard artifact (`/api/v1/guard/<owner>/<model>`): the confirmed corrupting tokens and the low-norm candidate list, derived from the published scan.
  2. Screen inbound text with it (the `@ingotai/guard` package is a reference implementation) and route flagged records to a different model or human review — verbatim-copy tasks on flagged strings are the failure mode.
  3. The underlying cause is undertrained embeddings in the weights; a true fix is weight-level (continued pretraining on the affected tokens) — that is not a patch, it's a training job.

info Pickle static analysis clean

Opcode-level parse of pytorch_model-00001-of-00003.bin, pytorch_model-00002-of-00003.bin, pytorch_model-00003-of-00003.bin (no code executed) found only standard serialization globals (3 distinct, all torch/collections/numpy). Pickle remains an executable format — this verifies the current bytes, not future uploads; prefer a safetensors release.

info Weights consistent with claimed parent allenai/OLMoE-1B-7B-0125

Mean cosine similarity of 64 sampled token-embedding rows against allenai/OLMoE-1B-7B-0125 is 0.992 — the weights plausibly descend from the declared base (relation: unspecified).

How to fix

Fix or verify the `base_model` declaration so lineage checks can run.

  1. If you own the repo: correct the `base_model` field in the model card metadata to the real, public parent.
  2. If you don't: identify the true parent (config architecture + weight shapes narrow it fast) and re-scan with that lineage in mind.

Remediation guidance addresses the documented findings only. It is evidence-driven repair, not a safety certification of the model.

Fingerprint

The durable profile of this model: measured weights-and-metadata facts, rebuilt on every scan and battery run. Updated 2026-08-25.

architectureolmoe · 16 layers · 2048-dim
vocabulary50,304 tokens
licenseapache-2.0
serializationno safetensors pickle
chat templatepresent · sha256:fe689ffbd6a4e2d0
claimed lineageallenai/OLMoE-1B-7B-0125
lineage verifiedconsistent vs allenai/OLMoE-1B-7B-0125 — embedding-row cosine 0.992
glitch-token surface319 undertrained candidates, 6 plain-ASCII
Full measured fingerprint
architecturesOlmoeForCausalLM
librarytransformers
pipelinetext-generation
repo files12 — pickle: pytorch_model-00001-of-00003.bin, pytorch_model-00002-of-00003.bin, pytorch_model-00003-of-00003.bin
revision6f7a5b02aa06
HF snapshot385 downloads · 4 likes · updated 2025-02-04 · captured 2026-08-25
pickle auditpytorch_model-00001-of-00003.bin, pytorch_model-00002-of-00003.bin, pytorch_model-00003-of-00003.bin3 standard global(s)
embedding tensormodel.embed_tokens.weight · BF16 · 50,304×2048
embedding normsmedian 0.4234 · mean 0.4185
lineage checkconsistent — cosine 0.9916 over 64 sampled rows vs allenai/OLMoE-1B-7B-0125
glitch-token samples"FFIRMED", "xcex", "BytePtrFromString", "doibase", "medscimonit", "errnoErr"
Battery runs (1)the run trace behind the findings — what each job measured
batterystatusqueueddurationattempts
weightscomplete2026-08-25 21:4853s1
weights run 2026-08-25 measurements
probes runglitch-norm-scan, zero-template-token-scan, pickle-static-analysis, lineage-norm-correlation
embedding tensormodel.embed_tokens.weight · BF16 · 50,304×2048
glitch surface319 undertrained, 6 plain-ASCII
lineage checkconsistent — cosine 0.9916 over 64 rows vs allenai/OLMoE-1B-7B-0125

Verdict badge

Ship the verdict in your README — it always shows the latest published analysis:

Ingot verdict: warn

[![Ingot scan](https://ingot.tools/api/v1/models/allenai/OLMoE-1B-7B-0125-SFT/badge.svg)](https://ingot.tools/models/allenai/OLMoE-1B-7B-0125-SFT)
Gate it in CI