Rebiha

Home › Datasets › Content & Marketing Copy

Content & Marketing Copy Fine-Tuning Dataset

Social posts, email campaigns, ads, and promotional content.

Fine-tune a model on this dataset

$29 for the dataset. Training is priced separately by model, starting at $47.

What is in this dataset

795 instruction-response pairs covering social posts, email campaigns, ads, and promotional content. Each pair is a realistic question with a full written answer, in the JSONL format used for supervised fine-tuning. No synthetic filler, no duplicated rows.

Example pairs

Real rows from the dataset, shown in full.

Write an Instagram caption for a product launch of a sustainable, refillable deodorant brand called 'FreshLoop'. The product is called 'Loop Stick'. Target audience: eco-conscious millennials and Gen Z. Tone: playful, inspiring, and planet-first. Include hashtags and a call to action.

Your armpits just got a major upgrade 🌱✨ Meet the Loop Stick – our first-ever refillable deodorant that’s as kind to your skin as it is to the planet. 🛸 Made with clean, aluminum-free ingredients and housed in a sleek, reusable case, it’s the deo that keeps on giving (literally – just pop in a new refill). Say goodbye to plastic waste and hello to all-day freshness that actually works. Because saving the planet shouldn’t mean sacrificing performance. 💚 Ready to loop in? Tap the link in bio to grab your Loop Stick starter kit. Let’s make every swipe count. 🌍 #FreshLoop #LoopStick #RefillableDeodorant #ZeroWasteBeauty #SustainableSwaps #DeodorantThatWorks

Write a promotional email for a 72-hour flash sale on a D2C mattress brand called 'Nod'. The sale offers 25% off all mattresses, plus a free pillow with every purchase. Subject line should be urgent but not spammy. Target: previous customers who haven't purchased in 6+ months.

Subject: 72 hours. 25% off. And a free pillow (just for you) ⏰ Hey [First Name], It’s been a while – and honestly, we’ve missed you. So here’s a little something to make up for lost time. For the next 72 hours only, we’re taking 25% off every single Nod mattress. That’s our biggest sale of the season – no code needed. 🛏️ And because we want your sleep to be *complete*, every mattress purchase comes with a free Nod cooling pillow (our bestseller, btw). Why now? Because you deserve better sleep tonight than you had last night. And because this deal disappears Sunday at midnight. 👉 [Shop the Flash Sale] Sweet dreams (finally), The Nod Team P.S. – This offer is for our returning…

Create a Facebook ad headline and body for a local coffee shop called 'Brew & Bloom' that just launched a subscription service for fresh-roasted coffee beans. Target audience: busy professionals in the local area (within 10 miles) who want premium coffee at home. Budget: small business, so keep copy concise and benefit-driven.

Headline: Fresh Roasted Coffee, Delivered to Your Doorstep – No Lines, No Waiting ☕ Body: Tired of running out of beans or settling for stale supermarket coffee? Brew & Bloom’s new subscription delivers our small-batch, locally roasted beans straight to your home – on your schedule. Choose your roast, frequency, and grind preference. First bag is on us. 🌿 👉 [Get Your First Bag Free] Limited to local delivery within 10 miles of downtown.

Who this is for

Teams generating social posts, ads, and email campaigns.

Which models this works with

The dataset is plain JSONL, so it works with any instruction-tuned open model. On Rebiha you can train it directly on:

Training uses LoRA, or QLoRA on a 4-bit base for larger models. Your base model's weights are never modified — training produces a separate adapter.

How it works

  1. Pick this dataset and a base model.
  2. Rebiha runs the job on a managed GPU. You set up nothing.
  3. Download your model: a ready-to-run GGUF, plus developer assets (adapter, tokenizer, configs).

Fine-tuning or RAG?

Use RAG when the model needs facts that change — your prices, your documents, today's inventory. Use fine-tuning when you want consistent style and structure without prompting for it every time. This dataset teaches the shape of a good answer in this domain, not facts to look up.

Related datasets

Start fine-tuning