Yixiang Gao

Could not detect voice activation whenever the signal start at the beginning

1

Hi, First, thank you for implementing this package. I was using it to per-process my voice signal but I encounter a problem with the signal particularly starts at the beginning....

Train Mask-RCNN

2

So far I am able to setup loading training data as well as iterate the `(image, target)` pair for running forward and backward. `MODEL=maskrcnn python3 examples/mlperf/model_train.py` `(, BoxList(num_boxes=50, image_width=1066, image_height=800,...

[WIP] merge_views fix proposal

1

Just wondering if I am on the right track, here is a test script that I used ``` from tinygrad.shape.shapetracker import ShapeTracker, View v1 = View((4,2), (2,1), 0, (1,1)) v2...

Not sure if what I did was the correct way. Had to delete the commits to sync on my fork, and then I pull to resolve some specific file issues....

Some investigation on why CUDA fail on CIFAR

1

In short, the block size it allocates exceeds the limit of the GPU (for mine 65535) ``` Device 0: "NVIDIA GeForce RTX 2070" CUDA Driver Version / Runtime Version 12.2...

While testing my fix for training CIFAR using CUDA, this error occurred after the latest merge.  something related to the fix in #1227

Fix cuda on cifar

4

Fix #1224 by adding `check_device_limit` If the allocated size is bigger than the limit, it creates additional cuda kernels with offset being manually added to the corresponding dimension index using...

include Zenoh info in README

2

Add the Zenoh middle ware notes as suggested.