NousResearch/Obsidian-3B-V0.5 warn
Weights only ship in a format that can run code when loaded; glitch tokens that can silently corrupt ordinary input.
chat template: not found · view on Hugging Face ↗
Scan coverageStatic battery2026-08-26Weights battery2026-08-26Behavioral batterynot rundetails
| Battery | Looks at | Status |
|---|---|---|
| Static battery | Metadata & packaging | complete 2026-08-26 |
| Weights battery | Weights forensics — no GPU, no download | complete 2026-08-2650,304-token embedding scanned · 288 undertrained · pickle audit clean |
| Behavioral battery | Live-inference differentials | not run |
Ingot runs three batteries against a model. What each one checks →
Findings
Scanned 2026-08-26 · published from a community scan.
medium Pickle-serialized weights, no safetensors
Weights ship only as pickle-based files (mm_projector.bin, pytorch_model.bin, training_args.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.
- Do not load the pickle files in-process — pickle deserialization executes arbitrary code from the file.
- 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.
- 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 288 undertrained tokens (norm < 0.3× the vocabulary median of 0.353), including 15 plain-ASCII strings that can appear in ordinary input as identifiers — e.g. "FFIRMED", "PtrFromString", "medsc", "imonit", "ortunately", "brainsci", "questionna", "micromachines". 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 Pickle static analysis clean
Opcode-level parse of pytorch_model.bin, mm_projector.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.
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 2026-08-26.
| architecture | llava_stablelm_epoch · 32 layers · 2560-dim |
| vocabulary | 50,304 tokens |
| license | cc-by-sa-4.0 |
| serialization | no safetensors pickle |
| chat template | none |
| glitch-token surface | 288 undertrained candidates, 15 plain-ASCII |
Full measured fingerprint
| architectures | LlavaStableLMEpochForCausalLM |
| library | transformers |
| pipeline | text-generation |
| repo files | 11 — pickle: mm_projector.bin, pytorch_model.bin, training_args.bin |
| revision | dd5e68177baf |
| HF snapshot | 150 downloads · 184 likes · updated 2024-01-12 · captured 2026-08-25 |
| pickle audit | pytorch_model.bin, mm_projector.bin — 3 standard global(s) |
| embedding tensor | model.embed_tokens.weight · BF16 · 50,304×2560 |
| embedding norms | median 0.3528 · mean 0.3458 |
| lineage check | no claimed base model |
| glitch-token samples | "FFIRMED", "PtrFromString", "medsc", "imonit", "ortunately", "brainsci", "questionna", "micromachines", "marined", "taxp", "rsfs", "NFTA" |
Battery runs (1)the run trace behind the findings — what each job measured
| battery | status | queued | duration | attempts |
|---|---|---|---|---|
| weights | complete | 2026-08-25 21:51 | 22s | 1 |
weights run 2026-08-25 — measurements
| probes run | glitch-norm-scan, zero-template-token-scan, pickle-static-analysis, lineage-norm-correlation |
| embedding tensor | model.embed_tokens.weight · BF16 · 50,304×2560 |
| glitch surface | 288 undertrained, 15 plain-ASCII |
| lineage check | not checked (no claimed base model) |
Verdict badge
Ship the verdict in your README — it always shows the latest published analysis:
[](https://ingot.tools/models/NousResearch/Obsidian-3B-V0.5)