TODO image

RewriteLikeMe-PeterThiel is a language model trained specifically to emulate the style of Peter Thiel. It can rewrite AI-written text to sound like Peter Thiel had written it. Due to the nature of its training data, opinion pieces posted on newspapers, this model is experienced in contrarian business and political commentary so it will probably produce better results if the input you give it is in the same general area as its training. Now, go forth and enjoy making text written like Peter Thiel!

Do you want to train AIs like this one, but on your own style? RewriteLikeMe.com (which I am working on) does just that! RewriteLikeMe also makes running models like this for writing much easier, and you can clone styles other than your own. The whole thing is very early beta, so please contact and shout at me if/when something explodes! I'd really appreciate your help in making this service work better!

Use this from https://rewritelikeme.com/download for a custom-built local chat inferface for rephrasing!

Also, there is a RewriteLikeMe Discord, for getting help and for sharing cool things that you generate or make with this tech. Check it out!

Included here are:

  • The LoRA files
  • The merged model files
  • The quantized merged model files (q8 GGUF) For usage with your preferred inference software.

Ideal sampling parameters:

temperature: 1.2,
top_p: 0.9,
min_p: 0.2
See model prompt (you really have to use this)

Put all this in the user message. The model is trained to rephrase text in the middle of a larger work without context, so even though it is single-turn, if the input is broken into chunks, you can rephrase texts of arbitrary length with this model.

You are an expert writing rephraser. Rephrase the TEXT TO TRANSFORM into something that possesses the same style AS IS DESCRIBED BY THE STYLE GUIDE. Do not change the content, just the style, of the text to transform. Sentence structure and some other things may be changed, even radically, so long as the meaning is not altered. Leave artifacts and seemingly cut-in-the-middle words at the start and end alone.

Note that <styleguide></styleguide> tags explain the target writing style. Analyze this reference to understand the specific tone, vocabulary, sentence structure, and stylistic nuances you should incorporate.

The style guide uses these headings with definitions to guide your rephrase:
- Structure (The Skeleton: How the Text is Built and Flows)
  Covers text architecture—bones and joints—for idea flow and engagement, including tempo/pacing, voice, emotion, formality, syntax/sentence variety, organization/flow.
- Lexicon (The Palette: Word Choices and Vocabulary)
  Involves word hue, texture, connotation for specificity, including diction, vocabulary range, repetition/variation, connotation/denotation.
- Rhythm and Sound (The Melody: Auditory Qualities)
  Sonic elements for memorability and cadence, including prosody, alliteration/assonance, meter/cadence, pause/silence.
- Rhetorical Devices (The Persuasion: Tools of Influence)
  Flourishes to argue, enchant, provoke; key for persuasion/art, including figurative language, emphasis, irony/contrast, directness vs. indirection.
- Tone and Mood (The Atmosphere: Emotional Resonance)
  Author's attitude (tone) and evoked feeling (mood) coloring the text, vital for emotion; includes attitudinal tone, evoked mood, humor/wit.
- Perspective and Narration (The Lens: Viewpoint and Focus)
  Positions reader/storyteller for subjectivity, including point of view, focalization, inclusivity.
- Imagery and Sensory Detail (The Texture: Vividness and Appeal)
  Sensory fabric for tangibility/evocation, including visual/sensory imagery, abstraction vs. concreteness.

Present your transformed text within <rephrase></rephrase> tags. Ensure that every piece of content from the original remains intact—only the stylistic presentation should change according to the style guide provided.

<styleguide>
{TODO FOR ME -- put this model's styleguide in here}
</styleguide>

<input>
{your AI-written text for rephrasing}
</input>
Downloads last month
31
GGUF
Model size
0.2B params
Architecture
llama
Hardware compatibility
Log In to view the estimation

8-bit

Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support