Independent publisher

Razavi

SaeedRazavi

Models in Library0

Datasets in Library1

Models on Hugging Face1

Followers—

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.

Published tasks: Text to image.

Reported licenses: other. Open a record to review its specific license and conditions.

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

Dataset · Text to image

blip3o-256-fid50k

Razavi

A frozen FID reference for text-to-image models trained on BLIP3o-256: 50k real (image, caption) pairs sampled from the training mixture (journeydb 5,903 / long-caption 37,518 / short-caption 6,579), per-shard reservoir sampling over 102 randomly chosen shards, seed 2026. Protocol: generate one image per caption in prompts/val/ (50k total), then compute FID against referencestats.npz. Only compare runs with the same prompt count. The shards are a subset of the BLIP3o pretraining data (BLIP3o/BLIP3o-Pretrain- on the Hugging Face Hub); see that release for the original licenses and terms.

Publicly accessible other

View dataset