mzient
Follow
Create your own GitHub profile
Sign up for your own profile on GitHub, the best place to host code, manage projects, and build software alongside 50 million developers.
Sign upPopular repositories
-
-
Forked from NVIDIA/DALI
A library containing both highly optimized building blocks and an execution engine for data pre-processing in deep learning applications
C++
-
-
Forked from pytorch/vision
Datasets, Transforms and Models specific to Computer Vision
Jupyter Notebook
465 contributions in the last year
Contribution activity
November 2020
Created 8 commits in 2 repositories
Created a pull request in NVIDIA/DALI that received 10 comments
Add support for UINT8 and INT8 outputs in CMN + scale and shift arguments
Signed-off-by: Michał Zientkiewicz mzient@gmail.com Why we need this PR? Pick one, remove the rest It adds new feature needed for inference in qua…
+155
−88
•
10
comments
Opened 6 other pull requests in 1 repository
NVIDIA/DALI
5
merged
1
open
Reviewed 30 pull requests in 1 repository
NVIDIA/DALI 30 pull requests
- Rework tutorials general
- NumpyReader : Replace std::regex with custom implementation
- Optimize the DCT GPU kernel.
- Ops rework to prepare iter-to-iter batch size variability
- Update DALI dependency
- COCOReader: Support for uncompressed RLE masks
- Update ExternalSource framework examples
- Compile DALI with Clang
- Move all NVTX infrastructure into core and create DALI domain
- MXNet Iterator: Revert to squeeze_labels=True behavior by default
- Hide hidden ops and improve Enum docs quality
- Example of random_mask_pixel to perform biased random crop
- Deprecate squeeze_labels option from MXNet iterator and enhance .squeeze function to match numpy style interface
- Enable compile time generation of dynlink wrappers for nvml
- Skips the building of prebuilt DALI package for nvidia-tensorflow
- COCOReader, parse RLE segmentation masks only when piwelwise masks output is requested
- Enables direct linking with libcuda.so instead of dlopen
- Add segmentation.RandomMaskPixel operator
- Fix build with all options off, relax libclang required version
- TensorVector update for iter-to-iter variable batch size
- Remove CUDA headers and generate stubs in runtime
- Reduce axes names
- One hot gpu
- Add SearchableRLEMask utility
- Make the custom nvJPEG allocator not throw and return only the status
- Some pull request reviews not shown.