​
🧠 Why This Matters

Don’t have a ready-made QA dataset? You don’t need one anymore. Whether you’re fine-tuning a model or building a knowledge assistant, having QA data is essential. But not everyone has the time or resources to create it from scratch. This feature lets you:
  • ✅ Convert any textual content into usable QA pairs.
  • 🚀 Instantly scale your training data, even from a few documents.
  • 🔄 Reuse your documentation, whitepapers, or reports to fuel model training.
  • 🛠️ Prepare data for model fine-tuning — natively supported in Prem Studio.

​
📁 What You’ll Need

To get started, you only need:
  • A set of text-based input files (e.g. .pdf, .txt, .docx, .html)
  • A Prem Studio account
  • Optionally: Custom instructions to guide the QA generation process
In this example, we’ll use 5 input files about crypto concepts and trends.

​
⚙️ Step-by-Step: Using QA Generation in Prem Studio

1

Create Dataset from the Datasets Section

GIF of navigating to synthetic dataFrom the left sidebar, click on Datasets, then + Create dataset (top right), and select the option Synthetic Data.
2

Upload Your Input Files

GIF of uploading input filesDrag and drop your documents (PDFs, DOCX, TXT, HTML). In our example, we use 5 crypto-related files.
3

(Optional) Add Custom Instructions

GIF of entering instructionsTailor the output with extra instructions, e.g.:
Only create questions related to blockchain, DeFi, and crypto tokens. Avoid generic or repetitive questions. Keep answers short and factual.
4

Generate Your Dataset

GIF of generating datasetClick Generate Dataset and let Prem Studio do the rest. In our example, we generate 100 QA pairs from 5 documents in under a minute.
5

Review Your Synthetic Dataset

GIF of reviewing and editing the datapointsClick Generate Dataset and let Prem Studio do the rest. In our example, we generate 100 QA pairs from 5 documents in under a minute.
Once the process is complete, you’ll be able to preview the generated dataset, edit any entries, and use it directly in Prem Studio’s fine-tuning and evaluation tools. Example Outputs:
QuestionAnswer
What is a blockchain?A decentralized, distributed ledger that records transactions.
What is the main purpose of tokens in DeFi?To facilitate decentralized financial operations.
How do smart contracts work?They execute predefined actions automatically based on conditions.

​
📦 What’s Next?

You can now:

​
💡 Pro Tips

  • Upload diverse content to get a wide range of QA pairs.
  • Use domain-specific instructions to control the output style.
  • Start small (5–10 files) to experiment, then scale up.