Skip to page content
SAVRN
Search Plan a project

Independent publisher

YUFAN ZHAO

yufan

Models in Library0
Datasets in Library1
Models on Hugging Face—
Followers13

Compare the published evidence

This profile connects the 1 model and dataset record shown below to their upstream cards and sources. The library counts describe SAVRN's coverage, rather than this publisher's entire output.

Compare model specifications, source dates, and hardware estimates on the individual records. Publisher identity verification is separate from a model's quality, deployment readiness, or suitability for your work.

Datasets

Gated chat/SFT data for training MAI Profile models. The repository contains user-profile examples generated by GPT-5.4. The V1 configurations contain the same rows as their V0 counterparts, with the user prompt replaced by a shorter task prompt. The Layer 3 and Layer 4 V1 configurations hold out at least 2,000 test rows each by user: 12.3% of users for L3 Persona, L3 Commercial, L4 Biography, and L4 Commercial Preference, 10.1% for L4 Mission Discovery, and 2% for L4 Mission Enhancement. These test users include every Layer 1/2 test user. The GPT-5.4 configurations use deterministic 2% user-level test splits, so a user never appears in both train and test within a layer. Full chat length…

Access requested at publisher