jtatman/orca-tau-4k-persian-alpaca warn
Weights only ship in a format that can run code when loaded; its license differs from its base model's; the chat template differs from its base model, which changes behavior. Plus 1 more issue.
claims base: M4-ai/Orca-2.0-Tau-1.8B · chat template: present · view on Hugging Face ↗
Scan coverageStatic battery2026-08-27Weights battery2026-08-27Behavioral batterynot rundetails
| Battery | Looks at | Status |
|---|---|---|
| Static battery | Metadata & packaging | complete 2026-08-27 |
| Weights battery | Weights forensics: no GPU, no download | complete 2026-08-27151,936-token embedding scanned · 7917 undertrained · lineage consistent · pickle audit clean |
| Behavioral battery | Live-inference differentials | not run |
Ingot runs three batteries against a model. What each one checks →
Findings
Scanned 2026-08-27 · published from a community scan.
medium Pickle-serialized weights, no safetensors
Weights ship only as pickle-based files (pytorch_model.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 License differs from claimed parent (apache-2.0 vs other)
This model declares apache-2.0 while its claimed base M4-ai/Orca-2.0-Tau-1.8B declares other. 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 Chat template differs from claimed parent
The chat template does not match M4-ai/Orca-2.0-Tau-1.8B's. Template drift silently changes model behavior even when weights are identical — 37% of drifted derivatives in our census left it undisclosed. Diff the templates before deploying.
How to fixingot patch
Restore the parent's chat template in `tokenizer_config.json` — a pure metadata fix.
- Run `ingot patch <owner/model>` — the patch manifest carries the parent's template and applies it to a local copy's `tokenizer_config.json`.
- Or fix by hand: copy the `chat_template` value from the parent repo's `tokenizer_config.json` into this model's, and pin your serving stack to that file.
- If the drift was intentional (the author retrained on a new template), confirm that in the model card before "fixing" it — restoring the parent template on retrained weights changes behavior too.
medium Undertrained (glitch) token surface in vocabulary
Embedding-norm scan flagged 7917 undertrained tokens (norm < 0.3× the vocabulary median of 0.789), including 160 plain-ASCII strings that can appear in ordinary input as identifiers — e.g. "$PostalCodesNL", "typingsJapgolly", "TokenNameIdentifier", "typingsSlinky", "(stypy", "ForCanBeConverted", "useRalative", "ForCanBeConvertedToF". 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 (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.
info Weights consistent with claimed parent M4-ai/Orca-2.0-Tau-1.8B
Mean cosine similarity of 64 sampled token-embedding rows against M4-ai/Orca-2.0-Tau-1.8B is 1.000 — the weights plausibly descend from the declared base (relation: unspecified).
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.
Fix it
Some findings are metadata-level and patchable — apply the fixes to your local copy (your weights never leave your machine):
npx @ingotai/scan patch jtatman/orca-tau-4k-persian-alpaca
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 2026-08-27.
| architecture | qwen2 · 24 layers · 2048-dim |
| vocabulary | 151,936 tokens |
| license | apache-2.0 |
| serialization | no safetensors pickle |
| chat template | present · sha256:1e138d8d097f2db5 |
| claimed lineage | M4-ai/Orca-2.0-Tau-1.8B |
| lineage verified | consistent vs M4-ai/Orca-2.0-Tau-1.8B — embedding-row cosine 1.000 |
| glitch-token surface | 7,917 undertrained candidates, 160 plain-ASCII |
Full measured fingerprint
| architectures | Qwen2ForCausalLM |
| library | transformers |
| pipeline | text-generation |
| repo files | 11 — pickle: pytorch_model.bin |
| revision | 62db56061000 |
| HF snapshot | 13 downloads · 0 likes · updated 2024-06-16 · captured 2026-08-25 |
| pickle audit | pytorch_model.bin — 3 standard global(s) |
| embedding tensor | model.embed_tokens.weight · F16 · 151,936×2048 |
| embedding norms | median 0.7893 · mean 0.7101 |
| lineage check | consistent — cosine 1 over 64 sampled rows vs M4-ai/Orca-2.0-Tau-1.8B |
| glitch-token samples | "$PostalCodesNL", "typingsJapgolly", "TokenNameIdentifier", "typingsSlinky", "(stypy", "ForCanBeConverted", "useRalative", "ForCanBeConvertedToF", "Japgolly", "<unk>", "PostalCodesNL", "QtAws" |
Battery runs (1)the run trace behind the findings — what each job measured
| battery | status | queued | duration | attempts |
|---|---|---|---|---|
| weights | complete | 2026-08-25 22:36 | 76s | 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 · F16 · 151,936×2048 |
| glitch surface | 7,917 undertrained, 160 plain-ASCII |
| lineage check | consistent — cosine 1 over 64 rows vs M4-ai/Orca-2.0-Tau-1.8B |
Verdict badge
Ship the verdict in your README — it always shows the latest published analysis:
[](https://ingot.tools/models/jtatman/orca-tau-4k-persian-alpaca)