Mosaic Composer

Hold

Tools

A training framework for building and running distributed deep learning workloads.

Why it's here

Placed in Hold: 1 article(s) of evidence from 1 source(s), led by research-stage coverage, with 0 in the last 30 days. Confidence 24%. Low accumulated evidence, so it defaults conservatively pending more signal.

Evidence (1)

4Hugging Face Blog·7/6/2026research
PRX Part 4: Data Strategy
Photoroom’s PRX team explains how it built the pre-training dataset for PRX using a mix of public and internal sources, then re-captioned images with a vision-language model. The article emphasizes broad coverage, long and accurate captions, light filtering, and packaging the corpus into a streamable format for distributed training.