Closed ma7dev closed 2 years ago
The ability to run approx.auto_cast_all() to auto-cast everything in the model (similar to what PyTorch has)
approx.auto_cast_all()
Let's actually drop this since it's unfeasible and has been replaced with the smarter auto_quantize feature in #16
Good point. I will close this issue.
The ability to run
approx.auto_cast_all()
to auto-cast everything in the model (similar to what PyTorch has)