README.md

---
license: mit
language:
- en
---
# Dataset Card for Daily Facts

It's now possible to edit facts directly into an LLM without retraining or finetuning. This dataset is a list of new facts generated by [exa.ai](https://exa.ai/)'s search 
to obtain new facts that can be fed into the model to tackle hallucination.

## Dataset Details

Inspired from [KnowEdit](https://huggingface.co/datasets/zjunlp/KnowEdit/viewer), the dataset has the following columns - "subject", "prompt", "ground_truth", "target_new". 


### Dataset Description

10 articles are chosen from exa's search API from each of the following domains - 
- "Latest news and advancements in Politics and facts",
- "Latest news in sport with facts",
- "Latest news in current events with facts"
- "Artificial intelligence advancements with facts",
- "Climate change impact",
- "Developments in quantum computing",
- "Recent breakthroughs in renewable energy",
- "Cryptocurrency market trends",
- "Space exploration and discoveries",
- "Global economic indicators",
- "Emerging medical treatments and research",
- "Cybersecurity threats and countermeasures",
- "Social media platform updates and policies",
- "Technological advancements in smartphones",
- "Environmental conservation efforts",
- "Developments in autonomous vehicles",
- "Breakthroughs in genomics and biotechnology

The articles are then curated into the following schema -
- "subject": "The subject of the fact that has its ground truth changing",
- "prompt": "A one line summary of the fact that is changing",
- "ground_truth": "What the previous value for the fact used to be.",
- "target_new": "What the new value for the fact will be now",

## Dataset Card Authors [optional]

[Shardul](https://github.com/bansal19), [Akshaj](https://github.com/aksj98) and [Shukant](https://github.com/ShukantPal)

For any questions, please reach out to Shardul directly at me@shardulb.xyz.