Tinytensor (wip)

• The Huggingface's transformers versions for inference under low resource ( no GPU )

Highlights

• Light installation footprint

• Blazing fast models ( majority models use less than 100M parameters )

• Relies on ONNX for weights inference (install size ~ 30MB)