THU-KEG/WildFB

WildFB Dataset WildFB (Wild Feedback) is a high-quality dataset of 186k instances filtered and refined from WildChat-4.8M. Each instance is labeled with a 4-level ordinal satisfaction score extracted from in-the-wild human-LLM interactions. Dataset Details WildFB addresses the challenge of training reward models without expensive human-annotated preference pairs. Instead, it extracts implicit reward signals from user follow-up queries in real-world conversations.… See the full description on the dataset page: https://huggingface.co/datasets/THU-KEG/WildFB.

<iframe src="https://silytics.si/embed/item/af3e73e4-d703-47f6-86af-a9e7db7fad67" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.