This is a Mistral-family model, not a Cerberus-uncensored category label.
annihilate_merge_metadata.json records the local annihilation/export path.
This repo is a full model export, not a PEFT/LoRA adapter.
The model uses the standard Mistral chat template and built-in Transformers MistralForCausalLM loader.

Quick Load

python

from transformers import AutoModelForCausalLM, AutoTokenizer

repo = "tjcrims0n/ms7"
tokenizer = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(repo, device_map="auto")

messages = [{"role": "user", "content": "Hello."}]
inputs = tokenizer.apply_chat_template(messages, return_tensors="pt", add_generation_prompt=True).to(model.device)
outputs = model.generate(inputs, max_new_tokens=80, eos_token_id=tokenizer.eos_token_id, pad_token_id=tokenizer.eos_token_id)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Smoke Test Notes

A local smoke test confirmed the repo loads with the built-in Mistral loader and produces coherent text. The arithmetic quirk 17 * 23 -> 381 also reproduces on the untouched cached base model, so it is not evidence of a broken annihilation export.