Thomas Viehmann
Thomas Viehmann
The inimitable @stas00 looked at Chapter 15 and found some errors/unclear parts. I'm collecting them here while I work on getting them into the errata. Thank you, Stas! Any remaining...
Dipping my toes into prims waters
Hi, thank you for unsloth. I happened to have triton 3.x dev installed, it seems the import location for libcuda_dirs changed. With this change, one can import unsloth with newer...
@kiya00 With any bit of luck, this should also unblock your enumerate PR.
This explores source locations for initial traces. The main alternatives I can see for making this a function call would be: - to set up a mapping at codegen and...
Some functions are actually OK, so we leave those alone. Fixes #95
Thanks for making LOEclipse! Here is something that tricked me more than a few times: When using my Python extension with LOEclipse, debugging (understandably) happens in some file under `dist/.ooo-debug`...
Hi, for fusing a not that much longer elementary pointwise computation, I get ``` with tvm::CompilationGroup_0 = graph(%0 : Float(*), %1 : Float(*)): %4 : int = prim::Constant[value=1]() %2 :...
We should look at how these are run. The numebrs are GB GPU memory, listing all that are > 0.6. (We could change to >0.5 as the threshold, but it...