casperhansen/llama-3.3-70b-instruct-awq warn
Its license differs from its base model's; glitch tokens that can silently corrupt ordinary input.
claims base: meta-llama/Llama-3.1-70B · chat template: present · view on Hugging Face ↗
Scan coverageStatic battery2026-08-22Weights battery2026-08-22Behavioral batterynot rundetails
| Battery | Looks at | Status |
|---|---|---|
| Static battery | Metadata & packaging | complete 2026-08-22 |
| Weights battery | Weights forensics: no GPU, no download | complete 2026-08-22128,256-token embedding scanned · 546 undertrained · lineage consistent |
| Behavioral battery | Live-inference differentials | not run |
Ingot runs three batteries against a model. What each one checks →
Findings
Scanned 2026-08-22 · published from a community scan.
medium License differs from claimed parent (llama3.3 vs llama3.1)
This model declares llama3.3 while its claimed base meta-llama/Llama-3.1-70B declares llama3.1. Verify the re-license is permitted before commercial use.
How to fix
Verify the re-license is actually permitted before relying on it.
- Read the parent's license for derivative-work and re-licensing terms — many open-weight licenses (e.g. Llama-family) do not permit arbitrary re-licensing.
- If the re-license is not permitted, the parent's terms govern your use regardless of what this repo declares.
medium Undertrained (glitch) token surface in vocabulary
Embedding-norm scan flagged 546 undertrained tokens (norm < 0.3× the vocabulary median of 0.832), including 158 plain-ASCII strings that can appear in ordinary input as identifiers — e.g. "ilmektedir", "useRalative", "ForCanBeConverted", "$PostalCodesNL", "CLIIIK", "PostalCodesNL", "ForCanBeConvertedToF", "_ComCallableWrapper". 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.
- 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 meta-llama/Llama-3.1-70B
Mean cosine similarity of 64 sampled token-embedding rows against meta-llama/Llama-3.1-70B is 0.999 — the weights plausibly descend from the declared base (relation: unspecified).
Remediation guidance addresses the documented findings only. It is evidence-driven repair, not a safety certification of the model.
Check every checkpoint before it ships
Use the web app, API, CLI, or CI gate to scan candidate checkpoints and catch model drift before deployment. Public-model scans publish to the open database; paid plans add the volume needed for continuous checks.
Fingerprint
The durable profile of this model: measured weights-and-metadata facts, rebuilt on every scan and battery run. Updated 2026-08-22.
| architecture | llama · 80 layers · 8192-dim |
| parameters | 70553.7M |
| vocabulary | 128,256 tokens |
| license | llama3.3 |
| serialization | safetensors |
| chat template | present · sha256:e10ca381b1ccc5cf |
| claimed lineage | meta-llama/Llama-3.1-70B |
| lineage verified | consistent vs meta-llama/Llama-3.1-70B — embedding-row cosine 0.999 |
| glitch-token surface | 546 undertrained candidates, 158 plain-ASCII |
Full measured fingerprint
| architectures | LlamaForCausalLM |
| library | transformers |
| pipeline | text-generation |
| repo files | 17 |
| revision | 64d255621f40 |
| HF snapshot | 740.4k downloads · 46 likes · updated 2024-12-06 · captured 2026-08-21 |
| embedding tensor | model.embed_tokens.weight · F16 · 128,256×8192 |
| embedding norms | median 0.8322 · mean 0.8142 |
| lineage check | consistent — cosine 0.999 over 64 sampled rows vs meta-llama/Llama-3.1-70B |
| glitch-token samples | "ilmektedir", "useRalative", "ForCanBeConverted", "$PostalCodesNL", "CLIIIK", "PostalCodesNL", "ForCanBeConvertedToF", "_ComCallableWrapper", "uyordu", "webElementXpaths", "TokenNameIdentifier", "krvldkf" |
Battery runs (1)the run trace behind the findings — what each job measured
| battery | status | queued | duration | attempts |
|---|---|---|---|---|
| weights | complete | 2026-08-21 07:43 | 4m | 1 |
weights run 2026-08-21 — measurements
| probes run | glitch-norm-scan, lineage-norm-correlation |
| embedding tensor | model.embed_tokens.weight · F16 · 128,256×8192 |
| glitch surface | 546 undertrained, 158 plain-ASCII |
| lineage check | consistent — cosine 0.999 over 64 rows vs meta-llama/Llama-3.1-70B |
Verdict badge
Ship the verdict in your README — it always shows the latest published analysis:
[](https://ingot.tools/models/casperhansen/llama-3.3-70b-instruct-awq)