jtatman/gpt2-open-instruct-v1-Anthropic-hh-rlhf warn
Weights only ship in a format that can run code when loaded. Plus 1 minor note.
claims base: vicgalle/gpt2-open-instruct-v1 · 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 2026-08-2650,260-token embedding scanned · 1 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 2026-08-26 · published from a community scan.
medium Pickle-serialized weights, no safetensors
Weights ship only as pickle-based files (checkpoint-22500/optimizer.pt, checkpoint-22500/pytorch_model.bin, checkpoint-22500/rng_state.pth, …). 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.
low Undertrained tokens in vocabulary (non-ASCII tail)
Embedding-norm scan flagged 1 undertrained tokens (norm < 0.3× the vocabulary median), but 0 decode to plain-ASCII strings, so exposure in English-language pipelines is limited. 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, checkpoint-22500/pytorch_model.bin, checkpoint-30000/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 vicgalle/gpt2-open-instruct-v1
Mean cosine similarity of 64 sampled token-embedding rows against vicgalle/gpt2-open-instruct-v1 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 profile of this model: measured weights-and-metadata facts, rebuilt on every scan and battery run. Updated 2026-08-26.
| architecture | gpt2 · 12 layers · 768-dim |
| vocabulary | 50,260 tokens |
| license | mit |
| serialization | no safetensors pickle |
| chat template | none |
| claimed lineage | vicgalle/gpt2-open-instruct-v1 |
| lineage verified | consistent vs vicgalle/gpt2-open-instruct-v1 — embedding-row cosine 1.000 |
| glitch-token surface | 1 undertrained candidates, 0 plain-ASCII |
Full measured fingerprint
| architectures | GPT2LMHeadModel |
| library | transformers |
| pipeline | text-generation |
| repo files | 40 — pickle: checkpoint-22500/optimizer.pt, checkpoint-22500/pytorch_model.bin, checkpoint-22500/rng_state.pth, checkpoint-22500/scheduler.pt, checkpoint-22500/training_args.bin, checkpoint-30000/optimizer.pt, checkpoint-30000/pytorch_model.bin, checkpoint-30000/rng_state.pth, checkpoint-30000/scheduler.pt, checkpoint-30000/training_args.bin |
| revision | 1259f032d4c1 |
| HF snapshot | 115 downloads · 5 likes · updated 2023-07-25 · captured 2026-08-25 |
| pickle audit | pytorch_model.bin, checkpoint-22500/pytorch_model.bin, checkpoint-30000/pytorch_model.bin — 3 standard global(s) |
| embedding tensor | transformer.wte.weight · F32 · 50,260×768 |
| embedding norms | median 3.9129 · mean 3.9278 |
| lineage check | consistent — cosine 1 over 64 sampled rows vs vicgalle/gpt2-open-instruct-v1 |
Battery runs (1)the run trace behind the findings — what each job measured
| battery | status | queued | duration | attempts |
|---|---|---|---|---|
| weights | complete | 2026-08-25 21:52 | 73s | 1 |
weights run 2026-08-25 — measurements
| probes run | glitch-norm-scan, zero-template-token-scan, pickle-static-analysis, lineage-norm-correlation |
| probes skipped | token-decode: no tokenizer.json |
| embedding tensor | transformer.wte.weight · F32 · 50,260×768 |
| glitch surface | 1 undertrained, 0 plain-ASCII |
| lineage check | consistent — cosine 1 over 64 rows vs vicgalle/gpt2-open-instruct-v1 |
Verdict badge
Ship the verdict in your README — it always shows the latest published analysis:
[](https://ingot.tools/models/jtatman/gpt2-open-instruct-v1-Anthropic-hh-rlhf)