Showcase

Despite being a tiny library, tinygrad is capable of doing a lot of things. From state-of-the-art vision to state-of-the-art language models.

Vision

You can either pass in the URL of a picture to discover what it is:

python3 examples/efficientnet.py ./test/models/efficientnet/Chicken.jpg

Or, if you have a camera and OpenCV installed, you can detect what is in front of you:

python3 examples/efficientnet.py webcam

Take a look at yolov8.py.

Take a look at whisper.py. You need pyaudio and torchaudio installed.

SMALL=1 python3 examples/whisper.py

python3 examples/stable_diffusion.py

"a horse sized cat eating a bagel"

You will need to download and put the weights into the weights/LLaMA directory, which may need to be created.

Then you can have a chat with Stacy:

python3 examples/llama.py

Make sure you have espeak installed and PHONEMIZER_ESPEAK_LIBRARY set.

Then you can talk to Stacy:

python3 examples/conversation.py