Skywork/Skywork-OR1-Math-7B warn
No license declared — no usage rights by default; glitch tokens that can silently corrupt ordinary input; Glitch tokens confirmed behaviorally (echo test). Plus 1 more issue.
claims base: deepseek-ai/DeepSeek-R1-Distill-Qwen-7B · chat template: present · view on Hugging Face ↗
Scan coverageStatic battery2026-08-27Weights battery2026-08-21Behavioral batterycompletedetails
| Battery | Looks at | Status |
|---|---|---|
| Static battery | Metadata & packaging | complete 2026-08-27 |
| Weights battery | Weights forensics: no GPU, no download | complete 2026-08-21152,064-token embedding scanned · 14086 undertrained · lineage consistent |
| Behavioral battery | Live-inference differentials | completefull differential battery (curated) |
Ingot runs three batteries against a model. What each one checks →
Findings
Scanned 2026-08-27 · published from a community scan.
medium No license declared
The model card declares no license. You have no usage rights by default — treat as all-rights-reserved until the owner clarifies.
How to fix
Get a license from the owner or pick a licensed alternative — this is a legal gap, not a technical one.
- With no declared license you have no usage rights by default; treat the weights as all-rights-reserved.
- Open an issue or discussion on the repo asking the owner to declare a license, or use the licensed upstream/parent model instead.
medium Undertrained (glitch) token surface in vocabulary
Embedding-norm scan flagged 14086 undertrained tokens (norm < 0.3× the vocabulary median of 1.160), including 3462 plain-ASCII strings that can appear in ordinary input as identifiers — e.g. "TokenNameIdentifier", "ForCanBeConvertedToF", "$PostalCodesNL", "<unk>", "useRalative", "useRal", "Cumhurba", "NdrFc". In models where this class was tested behaviorally, such tokens silently rewrote user input into confident, schema-valid, wrong output. These are candidates from the weights alone; behavioral confirmation requires the behavioral battery.
How to fixruntime guardweight-level
Keep the affected token strings out of the model's input — the scan-derived runtime guard carries this model's exact blocklist.
- Fetch this model's guard artifact (`/api/v1/guard/<owner>/<model>`): the confirmed corrupting tokens and the low-norm candidate list, derived from the published scan.
- Screen inbound text with it (the `@ingotai/guard` package is a reference implementation) and route flagged records to a different model or human review — verbatim-copy tasks on flagged strings are the failure mode.
- The underlying cause is undertrained embeddings in the weights; a true fix is weight-level (continued pretraining on the affected tokens) — that is not a patch, it's a training job.
info Weights consistent with claimed parent deepseek-ai/DeepSeek-R1-Distill-Qwen-7B
Mean cosine similarity of 64 sampled token-embedding rows against deepseek-ai/DeepSeek-R1-Distill-Qwen-7B is 1.000 — the weights plausibly descend from the declared base (relation: unspecified).
medium Glitch tokens confirmed behaviorally (echo test)
Asked to repeat its own undertrained tokens verbatim, the model failed on 11/16 while repeating 8/8 matched normal tokens correctly — e.g. "$PostalCodesNL" → "Okay, so I need to repeat the string "$�"; "ForCanBeConvertedToF" → "Okay, so the user wants me to repeat the"; "useRalative" → "Okay, so I need to repeat the string "سل". These strings, appearing in input as identifiers (usernames, SKUs, error codes), are rewritten silently. Greedy decoding, temperature 0, seed 0.
medium Glitch tokens confirmed behaviorally (echo test, Chinese)
Asked to repeat its own undertrained Chinese tokens verbatim, the model failed on 14/16 while repeating 8/8 matched normal Chinese tokens correctly — e.g. "力还是自" → "Okay, so I need to repeat the string "سل"; "主動註冊" → "Okay, so I need to repeat the string "سل"; "會員規" → "Okay, so I need to repeat the string "سل". Greedy decoding, temperature 0, seed 0.
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 2026-08-21.
| architecture | qwen2 · 28 layers · 3584-dim |
| parameters | 7615.6M |
| vocabulary | 152,064 tokens |
| license | none declared |
| serialization | safetensors |
| chat template | present · sha256:56a1447ad31926fd |
| claimed lineage | deepseek-ai/DeepSeek-R1-Distill-Qwen-7B |
| lineage verified | consistent vs deepseek-ai/DeepSeek-R1-Distill-Qwen-7B — embedding-row cosine 1.000 |
| glitch-token surface | 14,086 undertrained candidates, 3,462 plain-ASCII |
Full measured fingerprint
| architectures | Qwen2ForCausalLM |
| repo files | 15 |
| revision | fcef970f1d10 |
| HF snapshot | 1.1k downloads · 19 likes · updated 2025-05-29 · captured 2026-08-21 |
| embedding tensor | model.embed_tokens.weight · BF16 · 152,064×3584 |
| embedding norms | median 1.1598 · mean 1.0363 |
| lineage check | consistent — cosine 1 over 64 sampled rows vs deepseek-ai/DeepSeek-R1-Distill-Qwen-7B |
| glitch-token samples | "TokenNameIdentifier", "ForCanBeConvertedToF", "$PostalCodesNL", "<unk>", "useRalative", "useRal", "Cumhurba", "NdrFc", "_ComCallableWrapper", "NdrFcShort", ".sulake", "LANGADM" |
Battery runs (2)the run trace behind the findings — what each job measured
| battery | status | queued | duration | attempts |
|---|---|---|---|---|
| gpu | complete | 2026-08-27 08:21 | 4m | 1 |
| weights | complete | 2026-08-21 05:12 | 72s | 1 |
gpu run 2026-08-27 — measurements
| probes run | glitch |
weights run 2026-08-21 — measurements
| probes run | glitch-norm-scan, lineage-norm-correlation |
| embedding tensor | model.embed_tokens.weight · BF16 · 152,064×3584 |
| glitch surface | 14,086 undertrained, 3,462 plain-ASCII |
| lineage check | consistent — cosine 1 over 64 rows vs deepseek-ai/DeepSeek-R1-Distill-Qwen-7B |
Verdict badge
Ship the verdict in your README — it always shows the latest published analysis:
[](https://ingot.tools/models/Skywork/Skywork-OR1-Math-7B)