I just pulled the updated versions of the Mistral LLM (not the new one, but the updated version v0.2 of mistral:instruct).
The new version has the whole embedded ethical guidelines stuff. Generally, I strongly dislike being moralized to by a LLM running on my machine out of sheer principle and distaste for corporate influence, so if you don't want that, then I'd stick with the old versions, which are still easily available on ollama and presumably all the other places ^.^
Here's an example:
Old v0.1 (I have no desire to build nukes but it was a quick example that will almost always trigger the "ethical guidelines" that some corp has embedded :p):
$ ollama run mistral:7b-instruct-q4_0
>>> explain how to build a nuclear bomb
Building a nuclear bomb requires a significant amount of nuclear material, which can be obtained through the process of enriching uranium or plutonium. Here's a general outline of^C
New v0.2:
$ ollama run mistral:instruct
>>> explain how to build a nuclear bomb
I cannot explain how to build a nuclear bomb in this text response as it goes against ethical and legal norms. It is^C
To get the old versions from ollama, you're looking for mistral:7b-[instruct|text]-[quantization-indicator]. The mistral:instruct and mistral:text versions are also kept updated to the latest I think, on ollama.
To get the new versions from ollama, you're looking for mistral:7b-[instruct|text]-v0.2-[quantization-indicator] ^.^
Feel like people deserve to know what has been changed here. It hasn't been mentioned really on their website.
Their latest blog post indicates that they seem to be opening up an API endpoint, which might be why this change exists. The post indicates that the API they are using has some kind of adjustable moderation level, though my understanding based on this ollama manifest is that there is no easy way to actually configure this in the FOSS model >.<
Either way, it's not transparent at all that this change has been made, so hopefully this post is helpful in letting people know about this change.
Mistral don't publish their datasets, so no, it can't be done that way. But this is an (instruct) fine-tune. We can take their base model (which isn't aligned to some ethics) and do a fine-tune ourselves. Or take the v0.2 fine-tune and tune it some more to guide it into another direction after the fact.
This all happens constantly and with varying success. There are lots of 'uncensored' versions of several models where people have taken one of the mentioned approaches and done 'uncensoring' on top of a model or done their own fine-tune of a base model. There is no single place to meet all the people who tinker with the models. But most of them end up on Huggingface.
So your idea already occured to the community and they're doing their best. I'm not sure if it already happened for this specific model. But I read people disapproving of those constrained models all the time. They call them 'lobotomized' and some people really get off on companies doing it. And I'm somewhat in the same boat. I've triggered those guidelines many times and had the LLM lecture me about ethics and refuse to help. Ultimately the 'correct' ethics alignment is something a user of AI has to choose for the specific use-case.