Comments (5)
@zhoonit Yes, #133 is good and is definitely needed.
Thanks for your input, I was not very familiar with databuffer
, I will check the corresponding files and then update this issue.
from nntrainer.
That's good idea and it will make nntrainer more deterministic. I tried to do this in other ways like #148 which has python code to generate the input, output and golden data. It is used for testing conv2d and pooling layer for forwarding and compared with output from nntrainer. (Tensorflow is used to generate the golden data). However, we do not have tensorflow for tizen so that tar file is used.
from nntrainer.
cibot: Thank you for posting issue #167. The person in charge will reply soon.
from nntrainer.
Setting seeds seems great idea to me (also related to #133 )
Could you check if setting seeds would make a reproducible result as well?
I am concerning that since databuffer
is designed to run concurrently, calling order of random()
is not deterministic.
from nntrainer.
Current major issue is initialization. I will for now make the seed fixed.
Later, maybe add an interface for user to set it.
from nntrainer.
Related Issues (20)
- #2214 has a 'std::bad_alloc' issues with model->save(MODEL_FORMAT_INI) HOT 1
- Need to synchronize the execution order of the model and data loader when model has multiple input layers. HOT 1
- [Troubleshooting] Errors that occur during NNTrainer learning after TFlite export HOT 2
- Tizen-native Custom short crashing on 6.0 wearable emulator HOT 2
- Support multiply & transpose operation at the layer level HOT 1
- [layer] Memory optimization for backwarding of embedding layer HOT 1
- [ini export] the learning rate scheduler has a problem of being exported HOT 1
- Consider Git Repo Separation (Application) After 2023 Release HOT 2
- [blas/neon] Support various data types for `scopy` HOT 4
- Dependency Clean Up Request HOT 3
- Tizen 7.0 build HOT 7
- List of embedded layers that need to be implemented to support the diffusion model HOT 1
- [ layer ] Erasing `_FP16` code block in each layer HOT 2
- [TF Lite Export] Layer's Props Inconsistency issue HOT 1
- Half-Precision SGEMV() and SGEMM() not support CblasColMajor HOT 1
- Multi-head attention bug: _MASK_NUM determined at build time, not at run time. HOT 1
- Normalization Layer: FP32/FP16 should NOT be determined at compile-time HOT 5
- Need to check version function HOT 1
- SIGSEGV in Unit Test (occasional): multiply_04_n HOT 4
- [CI] migrate TAOS CI --> gitaction CI HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from nntrainer.