Skip to content

feat - add a vision at a clip universe subfolder #23

@rhysdg

Description

@rhysdg
  • House a number of examples that don't necessarily fit in the pure onnxruntime setting
  • Llava-next with mistral 7b for instance works really well on a single GPU, 4-bit quant, with ray serve and huggingface - adding example shortly. Noting that really well means it fits on a single GPU, it's still chunky so high latency on a consumer GPu at least

Metadata

Metadata

Assignees

Labels

enhancementNew feature or request

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions