AI & ML interests

One API for image, video, audio and language models: browser demos, datasets and API guides.

Recent Activity

stimiĀ  updated a Space 2 days ago
apimodels/README
stimiĀ  published a Space 2 days ago
apimodels/README
stimiĀ  updated a Space 2 days ago
apimodels/nemotron-transcribe
View all activity

stimiĀ 
updated a Space 2 days ago
stimiĀ 
published a Space 2 days ago
stimiĀ 
posted an update 2 days ago
view post
Post
34
We drew 36 Chinese idioms exactly as their four characters read, word by word, with Nano Banana 2.1, to make a picture-guessing game.

What made it work: never put the idiom itself in the prompt, write each literal scene by hand, and share one style text across all 36 that bans text and seal stamps. 36 images cost about $0.79 and most came back within 30 seconds.

The images, prompts, pinyin and English meanings are now a dataset (CC BY 4.0): apimodels/chinese-idioms-literal-illustrations

Write-up with the prompts that worked and the two that didn't: https://huggingface.co/blog/stimi/drawing-chinese-idioms-literally