-
I noticed that the output from my current download of nnef tools, using the caffe-to-nnef conversion with the python, results in a 512 byte header, in other words zero padding out to 0x200, while prev…
-
Created from https://github.com/caffe2/models/tree/master/googlenet
Commit: 65b8ec4b1eccfbce30e6ab70492c674aafbe007c
Command used to convert the model: python -m nnef_converter.convert --input-frame…
-
The Netron graph display tool is very useful, and they support many formats. It appears to me that if NNEF had a protobuf or flatbuffer format then it would not be a big issue to request that they a…
-
In this [functions](https://github.com/KhronosGroup/NNEF-Tools/blob/e84c3db29c1bffbd1938d40a10765badc0848606/parser/cpp/common/typespec.h#L275) static map with value type of `const Type*` is created. …
asdor updated
5 years ago
-
Let me reference @ajtulloch 's [comment](https://github.com/dmlc/tvm/pull/2116#issuecomment-444694200) about quantization workflow firstly:
> 1. Implement a model in a standard ML framework, genera…
-
I have a few questions about unpacking tuple results. On p. 23 it says "fragments that have multiple results essentially return a tuple."
First, this quote appears in the context of extended syntax…
zoeoz updated
5 years ago
-
The spec has some fragment examples that show only use of KHR_enable_fragment_definitions.
I found it necessary to use both extensions:
> extension KHR_enable_fragment_definitions;
extension KH…
-
I use `TensorHeader` struct to parse dat-files of tensor weights coefficients.
Its definition is in file `parser/cpp/common/binary.h`.
This is my code to write tensor weights:
```c++
std::ofstre…
igsha updated
6 years ago
-
value.h has
typedef float scalar_t;
Wouldn't it be preferable to keep everything double while expressions are being evaluated, and let the user round to float as needed?
The reason I'm bringing…
-
This page has a several tar'd up training models and eval models for tensorflow in protobuf, tflite and pbtext forms. They also have saved checkpoint data. These seem closer to what was used for c…