jhangmez/CHATPRG-v1.2-Phi-3.5-mini-instruct-GGUF warn

Convert the weights to safetensors before loading them anywhere that matters.

1. Do not load the pickle files in-process — pickle deserialization executes arbitrary code from the file.
2. 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.
3. Pin the exact revision hash you converted from, and load only your converted safetensors artifact from then on.

Align the stop-token declarations — a pure metadata fix to `generation_config.json` (and `config.json`).

1. 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.
2. Keep `config.json`'s `eos_token_id` consistent with (or a subset of) `generation_config.json`'s — runtimes differ in which file they read.
3. For the pad-equals-EOS hazard: repoint `pad_token` at a dedicated padding token before fine-tuning; serving is unaffected.
4. 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`).

Verify the re-license is actually permitted before relying on it.

1. 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.
2. If the re-license is not permitted, the parent's terms govern your use regardless of what this repo declares.

Restore the parent's chat template in `tokenizer_config.json` — a pure metadata fix.

1. Run `ingot patch <owner/model>` — the patch manifest carries the parent's template and applies it to a local copy's `tokenizer_config.json`.
2. 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.
3. 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.

Keep the affected token strings out of the model's input — the scan-derived runtime guard carries this model's exact blocklist.

1. 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.
2. 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.
3. 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.

Fix the GGUF's embedded metadata in place with gguf-py — template, EOS id, and pre-tokenizer are all metadata-editable; no requant needed.

1. 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.
2. Missing pre-tokenizer type: reconvert with a current `convert_hf_to_gguf.py`, or set the correct `tokenizer.ggml.pre` for the architecture.
3. 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>`.
4. 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.