nvidia/Qwen3.6-35B-A3B-NVFP4 warn
claims base: Qwen/Qwen3.6-35B-A3B · chat template: present · 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 248320 tokens (BF16, 2048-dim) found 1116 undertrained candidates, 341 plain-ASCII. Lineage vs Qwen/Qwen3.6-35B-A3B: consistent. Scanned 2026-08-20 (published from a community scan).
medium Undertrained (glitch) token surface in vocabulary
Embedding-norm scan flagged 1116 undertrained tokens (norm < 0.3× the vocabulary median of 0.567), including 341 plain-ASCII strings that can appear in ordinary input as identifiers — e.g. "tedothi", "ForCanBeConvertedToF", "ForCanBeConverted", "szexf", "Kinhted", "xfabl", "PostalCodesNL", "useRalative". 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 GPU deep 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 Qwen/Qwen3.6-35B-A3B
Mean cosine similarity of 64 sampled token-embedding rows against Qwen/Qwen3.6-35B-A3B is 1.000 — 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 | qwen3_5_moe |
| parameters | 18683.9M |
| vocabulary | 248,320 tokens |
| license | apache-2.0 |
| serialization | safetensors |
| chat template | present · sha256:e84f32a23fdda276 |
| claimed lineage | Qwen/Qwen3.6-35B-A3B |
| lineage verified | consistent vs Qwen/Qwen3.6-35B-A3B — embedding-row cosine 1.000 |
| glitch-token surface | 1,116 undertrained candidates, 341 plain-ASCII |
Full fingerprint
| architectures | Qwen3_5MoeForConditionalGeneration |
| library | Model Optimizer |
| pipeline | text-generation |
| repo files | 17 |
| revision | 491c2f1ea524 |
| HF snapshot | 12.3M downloads · 558 likes · updated 2026-06-12 · captured 2026-08-20 |
| embedding tensor | model.language_model.embed_tokens.weight · BF16 · 248,320×2048 |
| embedding norms | median 0.5666 · mean 0.5547 |
| lineage check | consistent — cosine 1 over 64 sampled rows vs Qwen/Qwen3.6-35B-A3B |
| glitch-token samples | "tedothi", "ForCanBeConvertedToF", "ForCanBeConverted", "szexf", "Kinhted", "xfabl", "PostalCodesNL", "useRalative", "tarsker", "useRal", "ejahter", "echslungs" |
Verdict badge
Ship the verdict in your README — it always shows the latest published analysis:
[](https://ingot.tools/models/nvidia/Qwen3.6-35B-A3B-NVFP4)