Ping Yu comments

Results 66 comments of


                                            Ping Yu

mapSync on Workers - and possibly on the main thread

We have learned from several important customers of TensorFlow.js, there is a strong need for synchronous reading API for WebGPU within web worker environment. I will list couple examples here:...

Error: The shape of dict[‘input_tensor’] provided in model.execute(dict) must be [1,-1,-1,3]

@gaikwadrahul8 I think you need to try the raw link of the model https://raw.githubusercontent.com/playground/tfjs-object-detection/main/model-json-service/model/model.json

[tfjs-converter] KeyError: 'kernel' when converting tfjs layers model to Keras

@TheCodez Layers model should be convertible to Keras model, can you share the model that cause the error? Also, I am curious what is the alternative way you succeeded in...

same input, devtools works perfectly, but in device did not work

The output looks like all the same value, can you verify if the inputs are correct on the device?

Output of MaxPooling2d not as expected on some Android devices with input tensor of certain shape.

This looks like it is overflowing, are you using float32 texture or float16 texture?

cannot import name 'shape_poly' from 'jax.experimental.jax2tf'

@gnecula we are having some issues with the CI, but we are expecting to have a new build in the new year.

Why the model predicted so slow on ios device?

@allrightzhu can you share the wechat app for blazeface?

Why the model predicted so slow on ios device?

@allrightzhu can you share it here? If you prefer more secure way, you can create a private github project and add me to the project. Thanks!

Why the model predicted so slow on ios device?

@allrightzhu I tried your facemesh demo on iphoneXR the console output indicates around 20ms, blazeface is slower at 30ms

Why the model predicted so slow on ios device?

@allrightzhu please let us know how did go, thanks.