Pundi AI Data Market – Complete Tutorial Guide

Pundi AI Data Marketplace
1

Pundi AI Data Market

Website: https://data.pundi.ai/market/

The Pundi AI Data Market is a decentralized marketplace where datasets are licensed via Access NFTs. These NFTs give you clear, verifiable rights to the dataset and can be used to train your LLMs or create dataset tokens (DTOKs) in the Data Pump.


Option A: Uploading a Dataset


1. Connect Your Wallet

  • Click “Connect” and choose your wallet (MetaMask, Pundi Wallet via WalletConnect, etc.).
  • Make sure you have $BNB for gas fees and $PUNDIAI (BEP-20) for transaction fees.
  • Approve the connection in your wallet.

2. Open “My Datasets”

  • At the top right, mouse over the menu (hamburger icon) next to your wallet address.
  • Select “My Datasets”.

3. Create a Dataset

  • Click “Create Dataset”.
  • Enter the dataset name.
  • Select the license type.
  • Set the visibility to “Public” or “Private”.

4. Create Dataset Card

  • Select the data type (text, images, audio, etc.).
  • Choose the language.
  • Select the list type.
  • Add data sources.

5. Add Description

  • In the README.md section, write a detailed description so users understand your dataset’s contents.

6. Finalize Dataset Creation

  • Click the blue “Create Dataset” button and approve in your wallet.

7. Upload Files

  • Drag and drop files into the upload area. Accepted formats: CSV, TXT, JSONL, ZIP, PARQUET.

8. Set Price

  • Set your dataset price in $PUNDIAI, $BNB, or $USDT.
  • Pay the fee in $PUNDIAI (BEP-20) with $BNB for gas fees.

9. Complete Listing

  • Once confirmed, your dataset will be listed on the Data Market.


Option B: Buying a Dataset

1. Connect Your Wallet

  • Click “Connect” and choose your wallet (MetaMask, Pundi Wallet via WalletConnect, etc.).
  • Make sure you have $BNB for gas fees and $PUNDIAI (BEP-20) for transaction fees.
  • Approve the connection in your wallet.

2. Browse the Data Market

  • Filter datasets by categories, tasks, languages, licenses, data type, row size, and format.

3. Select a Dataset

  • Click a dataset card to view details, file list, and README.md description.

4. Purchase

  • Click the “Buy” button.
  • Check the listed price ($PUNDIAI, $BNB, or $USDT) and confirm.
  • Approve the transaction in your wallet.

5. Access NFT Delivery

  • Once confirmed, you’ll receive an Access NFT giving you rights to use the dataset, including training your LLMs or creating a DTOK in Data Pump.


Video Tutorials

Pundi AI Data Market Playlist:
https://youtube.com/playlist?list=PLoFYck2JMxBZop2E8Eqrp7wwP3KPUrWum&feature=shared


GemGemCrypto Demo
https://youtu.be/GoRr9-CFR-w?feature=shared



Reference: Pundi AI Data Pump – Complete Tutorial Guide

Pundi AI Data Pump Playlist:
https://youtube.com/playlist?list=PLoFYck2JMxBYQ4netR-ZN8cMQso0Lzx2X&feature=shared



Disclaimer

This guide is for educational purposes only and does not constitute financial advice. Cryptocurrency trading and token creation involve risk. Always do your own research before making any purchase, sale, or token launch.

2 Likes
2

Ok I have come to understand getting points, making a Data set token , creating a data Pump DTOK, now I would like to understand better not so much how to make a data set , but how to create the contents of a data set properly, I winged it with some just by making a simple txt file, and a Zip on my other of the actual data product I attached, but I’m still purely guessing what i’m doing, I would like to see content context examples, of the makeup to a Data set truly desirable to the AI market place. For the data market not Meme DTOKs

3

Good question. For the Data Market, it’s less about just uploading files and more about making them structured and useful for training.

What works best:
Use CSV or JSONL so each row = one data point.
Keep it clean, consistent, and labeled (ex: text, sentiment, date).
Add context that saves AI devs time: tags, categories, metadata.

Examples buyers look for:
Text with labels (reviews + sentiment, Q&A pairs).
Images with annotations (objects, categories).
Audio with transcripts or emotion tags.
Niche sets (slang, domain-specific docs).

Structured and labeled data is what attracts real demand, while raw dumps are rarely useful on their own.

Here are two simple examples you can model from:

  1. Text dataset (CSV format)
text,sentiment,date
"This phone is amazing, battery lasts forever",positive,2025-08-20
"Customer service was rude and unhelpful",negative,2025-08-19
"Delivery was on time, product as expected",neutral,2025-08-18
  1. JSONL format (good for AI fine-tuning)
{"prompt": "Translate this to French: Hello, how are you?", "completion": "Bonjour, comment ça va ?"}
{"prompt": "Classify sentiment: I love this movie", "completion": "positive"}
{"prompt": "Summarize: The project deadline is next week", "completion": "Deadline next week"}

Notice:
Each row is one clear data point.
Fields are consistent (no random mixing).
Labels add context AI devs need.

You can also refer to some of the samples listed inside Data Market.

https://data.pundi.ai/market/repository/0xd6F81B022082700Ad1544F7fEa34C92DeE25B346/99903/

https://data.pundi.ai/market/repository/0xd6F81B022082700Ad1544F7fEa34C92DeE25B346/99626/

https://data.pundi.ai/market/repository/0xd6F81B022082700Ad1544F7fEa34C92DeE25B346/99899/

As the creator of your dataset, your wallet holds the Access NFT, which lets you add or remove files anytime. This keeps your dataset current and ensures it always has the latest information for training.

Dataset buyers hold the Access NFT, which gives them the right to download your files. Over time, this keeps their LLM training in sync with your latest data.

Hope the above brief info helps!

1 Like