llama.cpp reports the same tensor-count error for unsupported model layouts and damaged GGUF files. Add a focused hint so operators can update the backend or verify the model without losing the upstream diagnostic.
Assisted-by: Codex:gpt-5.6
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
