utensor_cgen

utensor_cgen copied to clipboard

Pre/Post Transform/Apply Observer

2

Implement an interface for users to register pre/post transformation/apply observers for `BackendPart` and `Transformer`. ```python @classmethod def regiseter_prior_observer(cls, callback): cls.prior_observer.append(callback) @classmethod def register_post_observer(cls, callback): # similar code def apply(self, ugraph):...

dboyliao

mbartling

Hyphens in TF operation names being written into generated C++ code

1

using : utensor-cli, version 0.3.5 I've just found that if Tensorflow operation names contain hyphens these are translated directly into c++ identifiers in the generated code, resulting in invalid code....

PeteBlackerThe3rd

Allow inline optimizer to spit out a single variable

4

To do weight updates it's much easier if only a single block of memory is responsible for all the weights. One way of doing this would be to concat all...

janjongboom

https://github.com/uTensor/utensor_cgen/blob/develop/utensor_cgen/backend/snippets/templates/containers/weight_header.hpp missing include guards similar to https://github.com/uTensor/utensor_cgen/blob/develop/utensor_cgen/backend/snippets/templates/snippets/get_ctx.hpp It's an easy fix and I'd normally get to it. For now, documenting

mbartling

High-Level Graph Builder

1

High level api for declaring IR-space graph. Use case: - pattern graph declaration for the graph matcher - user-defined graph and code generation.

dboyliao

Attribute error module has no attribute

2

I installed with pip but I am getting below Error  I tried with upgrade protobuf . If i tried with specified .whl then i am getting an import error....

saranworldrecord

A more fool-proof frontend parser selection scheme

1

Current frontend parser is selected via given graph file extension (such as `.pb` for an `GraphDefParser`...etc) @mbartling and @neil-tan are talking about a more fool-proof scheme at PR #49. Planning...

dboyliao

How to re-use a WrappedRamTensor and provide new input data

8

Hi, I am beginner with uTensor and embedded C/C++. I have a little experience around Python and wanted to study development of intelligence at the edge by building models in...

RajeshSiraskar

'TypeError: __init__() missing 1 required positional argument: 'output_nodes''

1

I got this message when trying to convert the simple_mnist.pb file using the command line: ``utensor-cli convert --output-nodes=y_pred simple_mnist.pb`` What is the correct way of running this conversion?

matheusataidee