uukuguy/speechless-coder-ds-6.7b warn
Weights only ship in a format that can run code when loaded; glitch tokens that can silently corrupt ordinary input; GGUF BPE tokenizer has no pre-tokenizer type. Plus 1 minor note.
chat template: not found · view on Hugging Face ↗
Scan coverageStatic battery2026-08-25Weights battery2026-08-25Behavioral batterynot rundetails
| Battery | Looks at | Status |
|---|---|---|
| Static battery | Metadata & packaging | complete 2026-08-25 |
| Weights battery | Weights forensics — no GPU, no download | complete 2026-08-2532,256-token embedding scanned · 1086 undertrained · pickle audit clean |
| Behavioral battery | Live-inference differentials | not run |
Ingot runs three batteries against a model. What each one checks →
Findings
Scanned 2026-08-25 · published from a community scan.
medium Pickle-serialized weights, no safetensors
Weights ship only as pickle-based files (pytorch_model-00001-of-00002.bin, pytorch_model-00002-of-00002.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.
low Padding token is the EOS token
The pad token and the (only) EOS token are the same. Fine-tuning frameworks mask pad positions out of the loss, so training on this checkpoint teaches the model to never emit EOS — the Phi-4 / Qwen 2.5 / DeepSeek R1 infinite-generation bug. Safe to serve, hazardous to fine-tune; repoint pad_token at a dedicated token first.
How to fixingot patch
Align the stop-token declarations — a pure metadata fix to `generation_config.json` (and `config.json`).
- Identify the token the chat template actually ends assistant turns with (e.g. `<|eot_id|>`, `<end_of_turn>`, `<|im_end|>`) and make sure its id is in `generation_config.json`'s `eos_token_id` list.
- Keep `config.json`'s `eos_token_id` consistent with (or a subset of) `generation_config.json`'s — runtimes differ in which file they read.
- For the pad-equals-EOS hazard: repoint `pad_token` at a dedicated padding token before fine-tuning; serving is unaffected.
- Until the repo is fixed, pass explicit stop tokens to your serving stack (e.g. vLLM `stop_token_ids`, llama.cpp `--override-kv tokenizer.ggml.eos_token_id`).
medium Undertrained (glitch) token surface in vocabulary
Embedding-norm scan flagged 1086 undertrained tokens (norm < 0.3× the vocabulary median of 8.875), including 343 plain-ASCII strings that can appear in ordinary input as identifiers — e.g. "milions", "anys", "desocupats", "persones", "unipersonals", "capbaix", "corresponia", "solteres". 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-00001-of-00002.bin, pytorch_model-00002-of-00002.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.
medium GGUF BPE tokenizer has no pre-tokenizer type
tokenizer.ggml.pre is missing, so llama.cpp silently falls back to the default pre-tokenization regex and splits text differently than the original model — degraded quality, worst on numbers, code, and non-English text (the pre-May-2024 conversion signature). Reconvert with a current converter, or set the correct tokenizer.ggml.pre with gguf-py; no requant needed.
How to fixingot patch
Fix the GGUF's embedded metadata in place with gguf-py — template, EOS id, and pre-tokenizer are all metadata-editable; no requant needed.
- Template or EOS drift: copy the current values from the source repo and write them into the GGUF (`gguf_set_metadata.py` / gguf-py) — the tensor data is untouched.
- Missing pre-tokenizer type: reconvert with a current `convert_hf_to_gguf.py`, or set the correct `tokenizer.ggml.pre` for the architecture.
- Until the file is fixed, override at load time: llama.cpp `--override-kv tokenizer.ggml.eos_token_id=int:<id>` and `--chat-template-file <fixed.jinja>`.
- Prefer a re-upload from the quantizer once the source repo's fix lands — already-downloaded GGUFs never pick up upstream fixes on their own.
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 uukuguy/speechless-coder-ds-6.7b
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 2026-08-25.
| architecture | llama · 32 layers · 4096-dim |
| vocabulary | 32,256 tokens |
| license | apache-2.0 |
| serialization | gguf pickle |
| chat template | none |
| glitch-token surface | 1,086 undertrained candidates, 343 plain-ASCII |
Full measured fingerprint
| architectures | LlamaForCausalLM |
| library | transformers |
| pipeline | text-generation |
| repo files | 13 — pickle: pytorch_model-00001-of-00002.bin, pytorch_model-00002-of-00002.bin |
| revision | c813a5268c6d |
| HF snapshot | 449 downloads · 7 likes · updated 2024-01-07 · captured 2026-08-25 |
| pickle audit | pytorch_model-00001-of-00002.bin, pytorch_model-00002-of-00002.bin — 3 standard global(s) |
| embedding tensor | model.embed_tokens.weight · F16 · 32,256×4096 |
| embedding norms | median 8.8749 · mean 8.5854 |
| lineage check | no claimed base model |
| glitch-token samples | "milions", "anys", "desocupats", "persones", "unipersonals", "capbaix", "corresponia", "solteres", "Naixements", "llogaters", "habitants", "unidenc" |
Battery runs (1)the run trace behind the findings — what each job measured
| battery | status | queued | duration | attempts |
|---|---|---|---|---|
| weights | complete | 2026-08-25 21:48 | 33s | 1 |
weights run 2026-08-25 — measurements
| probes run | glitch-norm-scan, zero-template-token-scan, pickle-static-analysis, lineage-norm-correlation, gguf-metadata |
| embedding tensor | model.embed_tokens.weight · F16 · 32,256×4096 |
| glitch surface | 1,086 undertrained, 343 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/uukuguy/speechless-coder-ds-6.7b)