Currently building Visual novel (untitled)
A visual novel in progress. Its art and video are generated with ComfyUI and open-source image and video models.
I build ML for chemistry & drug problems and games with generative AI, and write about both in a monthly journal.
A visual novel in progress. Its art and video are generated with ComfyUI and open-source image and video models.
GitHub
A BERT + ResNet-50 model that reads a post's text and image together to flag fake news.
Streamlit
Pulls reactions out of PDFs, converts them to SMILES and scores them on PMI, atom economy, E-factor, EcoScale and GHS hazards.
GitHub
A fine-tune of XTTS-v2 for Hindi speech on about 10 hours of audio, lowering eval loss by 12.7%.
Devlog
A first entry for the visual novel devlog, and a note on what this journal is for.
In progress
A visual novel in progress. Its art and video are generated with ComfyUI and open-source image and video models.
Publication
A critical review of carbon capture, utilisation and storage (CCUS) and the part it can play in a low-carbon future.