Tobias Alonso

icon indicating a website link https://www.linkedin.com/in/tobias-alonso/

icon company AMD icon location PhD in computer science and telecommunications. Currently working at @AMD.

Results 3 comments of Tobias Alonso

ModelWrapper.transform() mode returning "model_was_changed" flag

My general goal was to have a general streamlining transformation. The number of iterations over the sequence of transforms can vary from one net to another. We may set the...

Caching support for long-running transformations

For my test case using NUM_DEFAULT_WORKERS = 30 and StreamingFCLayer_Batch with "mem_mode" = "decoupled", these are the ones that take more time and I think are well suited for cache...

Detect and remove dead channels

In the same line, an approach that can provide additional optimization, possibly for all layers, is to keep track of the min and max value of each tensor. Currently, each...