google/gemma-4-31B-it pass
claims base: google/gemma-4-31B · chat template: not in config · view on Hugging Face ↗
Ingot findings
Static battery clean: safetensors weights, license declared, no template/tokenizer drift detected. Deep battery not yet run. Weights battery: embedding-norm scan over 262144 tokens (BF16, 5376-dim) found 0 undertrained candidates, 0 plain-ASCII. Lineage vs google/gemma-4-31B: consistent. Scanned 2026-08-20 (published from a community scan).
info Embedding-norm glitch scan clean
No undertrained tokens found: every non-special token's embedding norm is above 0.3× the vocabulary median (1.119). The glitch-token data-corruption class has no candidate surface in this model.
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.
- 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.
- 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.
- 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 Weights consistent with claimed parent google/gemma-4-31B
Mean cosine similarity of 64 sampled token-embedding rows against google/gemma-4-31B is 0.867 — 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.
- If you own the repo: correct the `base_model` field in the model card metadata to the real, public parent.
- 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 weights-and-metadata profile of this model, rebuilt on every scan and deep-battery run. Updated 2026-08-20.
| architecture | gemma4 |
| parameters | 31273.1M |
| vocabulary | 262,144 tokens |
| license | apache-2.0 |
| serialization | safetensors |
| chat template | none |
| claimed lineage | google/gemma-4-31B |
| lineage verified | consistent vs google/gemma-4-31B — embedding-row cosine 0.867 |
| glitch-token surface | clean no undertrained tokens |
Full fingerprint
| architectures | Gemma4ForConditionalGeneration |
| library | transformers |
| pipeline | image-text-to-text |
| repo files | 12 |
| revision | 842da3794eaa |
| HF snapshot | 9.3M downloads · 3.6k likes · updated 2026-07-20 · captured 2026-08-20 |
| embedding tensor | model.language_model.embed_tokens.weight · BF16 · 262,144×5376 |
| embedding norms | median 1.1186 · mean 1.1182 |
| lineage check | consistent — cosine 0.8668 over 64 sampled rows vs google/gemma-4-31B |
Verdict badge
Ship the verdict in your README — it always shows the latest published analysis:
[](https://ingot.tools/models/google/gemma-4-31B-it)