01-ai/Yi-1.5-9B-Chat warn
Chat template ends turns with <|im_end|>, which is not a configured stop token; glitch tokens that can silently corrupt ordinary input. Plus 1 minor note.
chat template: present · view on Hugging Face ↗
Scan coverageStatic battery2026-08-27Weights battery2026-08-25Behavioral batterycompletedetails
| Battery | Looks at | Status |
|---|---|---|
| Static battery | Metadata & packaging | complete 2026-08-27 |
| Weights battery | Weights forensics: no GPU, no download | complete 2026-08-2564,000-token embedding scanned · 392 undertrained |
| Behavioral battery | Live-inference differentials | completefull differential battery (curated) |
Ingot runs three batteries against a model. What each one checks →
Findings
Scanned 2026-08-27 · published from a community scan.
medium Chat template ends turns with <|im_end|>, which is not a configured stop token
The chat template terminates assistant turns with <|im_end|>, but the effective EOS set (config.json ∪ generation_config.json = [2] → ["<|endoftext|>"]) never stops on it. Config-honoring runtimes generate past the terminator until the token budget is exhausted — runaway cost and self-continuing fake turns. Add <|im_end|>'s id to generation_config.json's eos_token_id.
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 392 undertrained tokens (norm < 0.3× the vocabulary median of 0.927), including 150 plain-ASCII strings that can appear in ordinary input as identifiers — e.g. "<0xFC>", "<0xFF>", "<0xFD>", "<0xFB>", "<0xFE>", "<0xFA>", "mabaochang", "nzoem". 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.
low Partial glitch-token echo degradation
Echo failures on 1/16 undertrained tokens vs 0/8 controls — a differential exists but below the confirmation bar (≥50% glitch failures with clean controls).
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 01-ai/Yi-1.5-9B-Chat
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 · 48 layers · 4096-dim |
| parameters | 8829.4M |
| vocabulary | 64,000 tokens |
| license | apache-2.0 |
| serialization | safetensors |
| chat template | present · sha256:91aa728ae59c8e30 |
| glitch-token surface | 392 undertrained candidates, 150 plain-ASCII |
Full measured fingerprint
| architectures | LlamaForCausalLM |
| library | transformers |
| pipeline | text-generation |
| repo files | 15 |
| revision | 1a0fc698cf88 |
| HF snapshot | 17.1k downloads · 149 likes · updated 2024-06-26 · captured 2026-08-25 |
| embedding tensor | model.embed_tokens.weight · BF16 · 64,000×4096 |
| embedding norms | median 0.9271 · mean 0.9024 |
| lineage check | no claimed base model |
| glitch-token samples | "<0xFC>", "<0xFF>", "<0xFD>", "<0xFB>", "<0xFE>", "<0xFA>", "mabaochang", "nzoem", "mcited", "mrrooter", "vepfs", "mrroot" |
Battery runs (3)the run trace behind the findings — what each job measured
| battery | status | queued | duration | attempts |
|---|---|---|---|---|
| gpu | complete | 2026-08-27 04:31 | 53s | 1 |
| gpu | complete | 2026-08-25 21:47 | 89s | 1 |
| weights | complete | 2026-08-25 19:23 | 23s | 1 |
gpu run 2026-08-27 — measurements
| probes run | glitch |
gpu run 2026-08-25 — measurements
| probes run | glitch |
weights run 2026-08-25 — measurements
| probes run | glitch-norm-scan, zero-template-token-scan, lineage-norm-correlation |
| embedding tensor | model.embed_tokens.weight · BF16 · 64,000×4096 |
| glitch surface | 392 undertrained, 150 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/01-ai/Yi-1.5-9B-Chat)