4 Commits
Author SHA1 Message Date
Ayush RanjanandgVisor bot 8c386259e0 Add /usr/local/nvidia/bin to PATH in ffmpeg image.
This image is being used for some GPU tests.

Also clean up /bin/nvidia/bin from PATH from all GPU images. It's not used.

PiperOrigin-RevId: 720649135
2025-01-28 11:35:48 -08:00
Etienne PerotandgVisor bot 3f058a8436 Harmonize ollama Docker image version.
The benchmark image was updated to 0.5.1, so this updates the other images
to the same version.

PiperOrigin-RevId: 706801311
2024-12-16 12:45:11 -08:00
Andrei VaginandgVisor bot b27d4b5834 images: don't build unused images
Runtime and gpu tests are running on x86_64 only.

PiperOrigin-RevId: 613771002
2024-03-07 18:40:14 -08:00
Etienne PerotandgVisor bot 35cb2282f0 Add a single-model ollama image.
Yes, it's yet another huge image (26GB) :)

This image is useful to benchmark checkpoint/restore performance.
To fairly measure it, we need an image that has only one model
preloaded into it, as this mirrors the situation where someone might
want to use checkpoint/restore for an inference model where there is
only one model to serve.

This image uses mixtral as a medium-large model. The larger the model,
the more important the gains in checkpoint/restore performance, so
measuring with a large model provides a more representative and
significant result.

PiperOrigin-RevId: 613335037
2024-03-06 14:10:26 -08:00