[Tensor] CharTensor class with signed 8-bit integer #2529
Conversation
djeong20
requested review from
myungjoo,
jijoongmoon,
again4you,
jaeyun-jung,
leemgs,
wooksong,
helloahn,
kparichay,
gichan-jang,
anyj0527,
zhoonit,
lhs8928,
songgot,
jihochu,
DonghakPark,
SeoHyungjun,
baek2sm,
skykongkong8,
EunjuYang and
a team
as code owners
|
📝 TAOS-CI Version: 1.5.20200925. Thank you for submitting PR #2529. Please a submit 1commit/1PR (one commit per one PR) policy to get comments quickly from reviewers. Your PR must pass all verificiation processes of cibot before starting a review process from reviewers. If you are new member to join this project, please read manuals in documentation folder and wiki page. In order to monitor a progress status of your PR in more detail, visit http://ci.nnstreamer.ai/. |
|
|
cibot: 
|
|
djeong20
force-pushed
the
add/char_tensor
branch
from
7bb62ef
to
50752a8
Compare
djeong20
force-pushed
the
add/char_tensor
branch
from
50752a8
to
dd3b891
Compare
djeong20
changed the title
djeong20
force-pushed
the
add/char_tensor
branch
from
dd3b891
to
cf08ee7
Compare
nntrainer/tensor/char_tensor.cpp
Outdated
| case Initializer::ONES: | ||
| setValue(1.0f); | ||
| break; | ||
| case Initializer::LECUN_NORMAL: |
There was a problem hiding this comment.
Could you check whether these initializers also work for int datatype?
If it is not, then it gives proper notification.
There was a problem hiding this comment.
currently, these initializers won't work for tensors with integer type and throw an error. I intentionally left the code for future work. however, if these initializers are not valid for integers, they can be removed such as follows.
| case Initializer::LECUN_NORMAL: | |
| case default: | |
| throw std::invalid_argument("Initializer not valid in current data type"); |
I'll do some research and update this in the later PR!
There was a problem hiding this comment.
I figured both normal and uniform are not possible for integer types, so I made the changes. PTAL!
In this PR, a new type of tensor, the CharTensor class, is designed explicitly for handling signed 8-bit integer data types that have already undergone quantization. This new tensor class aims to provide users with more options when working with tensors and their respective data types. Currently, the CharTensor class does not support mathematical operations like multiplication or addition. However, these features will be added in future updates. **Self-evaluation:** 1. Build test: [X]Passed [ ]Failed [ ]Skipped 2. Run test: [X]Passed [ ]Failed [ ]Skipped Signed-off-by: Donghyeon Jeong <dhyeon.jeong@samsung.com>
djeong20
force-pushed
the
add/char_tensor
branch
from
cf08ee7
to
50cfa00
Compare
In this PR, a new type of tensor, the CharTensor class, is designed explicitly for handling signed 8-bit integer data types that have already undergone quantization.
This new tensor class aims to provide users with more options when working with tensors and their respective data types.
Currently, the CharTensor class does not support mathematical operations like multiplication or addition. However, these features will be added in future updates.
Self-evaluation: