Training a model on my content

Beginners
1

Hi guys (any advice pls!)

I have 80 blogs which I need to re-purpose into social media content.

Posts need to be using images in the blogs, with hyperlinks to sources, within social media character limit constraints, and in my brand’s specific voice and style.

How would I go about getting a model trained on my existing content to be able to massively increase my content creation capacity?

2

80 blogs is enough for fine-tuning a small model on your brand voice, but the real bottleneck won’t be the generation — it’ll be maintaining quality at scale. A few practical steps:

  1. Fine-tune a small LLM (Mistral 7B or Llama 3 via LoRA) on your 80 blogs as context + your existing social posts as target outputs. Focus on tone, structure, length constraints.
  2. Prompt engineering first — before training, try few-shot prompting with your best 5-10 blog-to-social examples. That alone might get you 80% of the way.
  3. Build a review loop — generate 5 variants per post, have a human pick the best. Use those picks as your training set for iteration.

Training on 80 samples won’t produce production-ready output on its own, but as a bootstrap for a human-in-the-loop process it works fine. Scale the dataset by keeping every approved generation.

3

Hy, I’ve made something similar: I’ve created a dataset here on hugging face uploading the full text and single chapter folder of my sci-fi novel “432 Un Grande Viaggio” (italian) and “432 A Journey Beyond”. Then I’ve used Claude Opus to help me write a 3000 rows python code to start a Hugging Face Space using a cascade of open source models (Qwen and others for fallback). Now the space is running and my characters can reply to user’s prompts with their own personalities , basing their answers on my novel’s dataset.
I think you could do something similar for your blogs: You’ll have a dataset containing your blog’s post and the system will do RAG on this for any answer. You can insert in the .py code all the instructions (e.g “write always with my own style, as you can read in the .txt provided”.. and so. You can start the model system prompt with something like "You are [your name], a blogger, a writer, writing about [arguments] . Your typical target persons are [specify] … " and so on. You can add whatever you need to the syestem prompt (fixed rules, etc. ) … My space is public , you can visit it here:

but if you need you can set your space to “Private” and only you and eventually authorized contributors will have access.