🧩 Training Files

In the Training Files section, you can upload documents or images that will be used to automatically create AI training datasets.
You can access this section here: https://ainisa.com/business/training-files

📁 Supported File Types

• PDF documents
• Image files: .png, .jpg, .jpeg

👉 Both text-based PDFs and scanned image PDFs can be read and parsed automatically by Ainisa.

---

🪜 Steps to Create Datasets from Files

1. Upload your file
   Click Upload Files (top-right button) and select a PDF or image. After uploading, the file will appear in the table.

2. If the file is a PDF

   • After upload, Ainisa starts parsing the PDF automatically.
   • The “Number of Pages” column shows how many pages are ready.
   • When all pages are processed, click the ⋯ icon and choose Create Question-Answer Dataset.

3. If the file is an image (jpg, jpeg, png)

   • Images are processed instantly.
   • Click the ⋯ icon → Create Question-Answer Dataset.

   

4. Parsing progress

   • For PDFs, the process may take some time depending on file size.
   • You can monitor progress under the Parse Stage column (0% → 100%).
   • Once it reaches 100%, all pages have been converted into datasets.

---

📊 Where do the datasets go?

After you click Create Question-Answer Dataset, Ainisa automatically generates Q&A items and saves them in the Datasets section:
https://ainisa.com/business/datasets

• If the dataset was created from a file, you’ll see the file name in the dataset list.
• If it was created manually, no file name will appear.

---

✅ Example Table

File Name	File Type	Number of Pages	Parse Status	Parse Stage
agentic-rag.jpeg	jpg	1/1	Parsed to image for reading	0%
Ainisa — partnership.pdf	pdf	8/8	Parsed to image for reading	0%

---

💡 Notes

• Use this section if you already have content (PDFs, scanned images, etc.) and want to auto-generate datasets instead of writing them manually.
• This step only prepares data — actual model training happens later in the Fine-tuning section.
• Each processing image or pdf consumes API tokens from your connected OpenAI account.
• There are limits for reading pages:
  • Each image counts as 1 page.
  • A 10-page PDF counts as 10 pages.
  • Each subscription plan has its own monthly page limit for dataset creation.
  • If the user reaches the page limit, the reading/parsing process will stop and fail.
  • Example: If the Pro Plan allows 100 pages/month and the user already used 70, uploading a 50-page PDF will only process 30 pages; the remaining 20 will fail.
  • To continue, users can upgrade to a higher plan or wait until the next subscription renewal date.