Taehee Jeong comments

Results 15 comments of


                                            Taehee Jeong

Mismatch in number of weights when loading quantized model (activation layer)

Hi @Lotte1990 , sorry for late response. Want to check if this still bugs you, before taking a look into this.

tf.quantization.quantize fails when converting to TFLite

Hi @leandro-gracia-gil , As you observed, `tf.quantization.quantize` is not converted to TFLite op. You can try [fake_quant_with_min_max_args](https://www.tensorflow.org/api_docs/python/tf/quantization/fake_quant_with_min_max_args) to actually quantize and dequantize the value, but I can't recall a op...

sparsity.prune_low_magnitude fails with mixed precision policy mixed_float16

Hi @dd1923 , sorry for really late response. Just want to check if this still bugs you before picking up this again.

Add context free proto file grammar

+1, Was there a reason this never got merged?

Integer quantization converts bias of 32-bit float type in conv2d to 8-bit int type on TFLite

This seems to be resolved with nightly version of tensorflow. Can you try again?