Munaza Ashraf

Munaza10

·

AI & ML interests

None yet

Recent Activity

repliedto bghira's post 6 days ago

I've been gradually recaptioning aged text-to-image datasets with better vision models! These datasets are also repackaged into the more modern webshart format (https://github.com/bghira/webshart) which includes built-in aspect bucketing and caption delivery. The first two datasets are ready for use! - https://huggingface.co/datasets/webshart/terminusresearch-photo-anatomy - https://huggingface.co/datasets/webshart/terminusresearch-photo-aesthetics "anatomy" is a bunch of human-centric images containing people holding or otherwise interacting with objects or positioned in complex ways. "aesthetics" is a collection of visually striking images - high contrast, diverse colouration, and cinematic framing (among other factors). These two datasets from 2023 were recaptioned with GLM 5.3 Flash via Ollama Cloud and Zhipu AI APIs, with a smaller portion run over 4x H100 with GLM 5.3 Flash in W4A16 precision - these outputs were checked by hand for quality, and the 4bit run was continued at a batch size of 64. What datasets would you like to see recaptioned next? A better CC12M is on its way!

reacted to bghira's post with 😎 6 days ago

I've been gradually recaptioning aged text-to-image datasets with better vision models! These datasets are also repackaged into the more modern webshart format (https://github.com/bghira/webshart) which includes built-in aspect bucketing and caption delivery. The first two datasets are ready for use! - https://huggingface.co/datasets/webshart/terminusresearch-photo-anatomy - https://huggingface.co/datasets/webshart/terminusresearch-photo-aesthetics "anatomy" is a bunch of human-centric images containing people holding or otherwise interacting with objects or positioned in complex ways. "aesthetics" is a collection of visually striking images - high contrast, diverse colouration, and cinematic framing (among other factors). These two datasets from 2023 were recaptioned with GLM 5.3 Flash via Ollama Cloud and Zhipu AI APIs, with a smaller portion run over 4x H100 with GLM 5.3 Flash in W4A16 precision - these outputs were checked by hand for quality, and the 4bit run was continued at a batch size of 64. What datasets would you like to see recaptioned next? A better CC12M is on its way!

updated a dataset about 2 months ago

Munaza10/llm-tarot-dataset

View all activity

Organizations

Collections 1

models 3

Munaza10/llm-tarot-dataset

Munaza10/ZimageLorawithnocaps

Text-to-Image • Updated Jun 19 • 7

Munaza10/ZImageLorawithcaps

Text-to-Image • Updated Jun 19 • 5

datasets 1

Munaza10/llm-tarot-dataset

Viewer • Updated Jul 23 • 78 • 222