Looking for a Local Model with Fine-Tuning and No Censorship

Beginners
1

Hello!

I am looking for a model for local use that meets the following requirements:

  1. Edit and generate responses based on information I provide during training or fine-tuning.
  2. Comes pre-trained on a large dataset of books and texts for a strong foundational use.
  3. Supports additional fine-tuning so I can integrate my own knowledge and datasets.

My goal is to create a virtual assistant that:

  • Allows me to add textual information, as well as images (e.g., diagrams, projects).
  • Ideally supports audio and video inputs. If that’s not possible, it should at least store references/links to external sources (e.g., YouTube channels or local folders).
  • Operates entirely locally, without sending any data to external servers.

Additional features I would like:

  1. The ability to define user roles:
  • Regular users with limited access.
  • Admin/root users with full access, secured by a password.
  1. The model should be uncensored and free from content restrictions.
  2. Full autonomy to ensure that all data remains private and is stored locally.

The ultimate purpose of this model is to create a personal archive that can be passed down through generations. For instance, I want my great-great-grandchildren, 200 years from now, to be able to “interact” with a version of me through this assistant.

Could you recommend a suitable model for this purpose (preferably in GGUF format for LM Studio) and provide guidance on where to download it? Thank you!

2 Likes
2

Why the model should be uncensored if it’s for a version of you (just asking not judging!) ?

Perhaps if you want say more about your needs, you will have more answers, even, I admit it, you provided a lot of details, but I feel like it is not enough more for gurus, here, answer properly to your request.

1 Like
3

Maybe he wasn’t it to be able to tell his great grandkids “Well my boy, in my day we knew the TV was on when the white power light was glowing.” without it kvetching that it can’t generate the response because it promotes violence. ChatGPT literally did this to me. :person_tipping_hand:

4

I want to include information including sexual implications, oriental medicine, and personal experience in developments. Therefore, I need the model to have no restrictions, since I believe that the restrictions set by other people do not agree with my views.

5

By the way, this often happens when someone adds to censorship a ban on, say, a solution to combating viruses, where they rely only on expert opinion, although I personally have experience in such matters and in the fight against many supposedly incurable diseases, including CANCER and AIDS , and much more