search

hikettei / cl-waffe2

[Experimental] Graph and Tensor Abstraction for Deep Learning all in Common Lisp
https://hikettei.github.io/cl-waffe2/
MIT License
122 stars 5 forks source link

Enhancement on reduction operations: !argmax/!argmin/!max/!min #60

Closed hikettei closed 11 months ago

hikettei commented 11 months ago

  1. Fixed the issue where !argmax/!argmin wasn't working well due to optimization of call-with-view. It was solved by adding a keyword argument :force-order, which will ignore flatten if t.

  2. Added CPUTensor Implementation of !argmax/!argmin/!max/!min, but it does not use SIMD. maybe someone has to write C kernel directly for the future release for more speeds.

  3. Added new nodes: MaxValue-Node, MinValue-Node.