denizyuret
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 up
Pinned
336 contributions in the last year
Contribution activity
July 2020
Created an issue in JuliaGPU/CUDA.jl that received 5 comments
Performance: sum
Regular sum is important when computing loss. The one with dims argument is important when computing bias backward pass. The one with a function ar…
5
comments
- Create Knet.Models20
- Create Knet.Layers20
- Create Knet.Ops20 and add benchmarking for operator set
- Test CuArrays on tutorial and examples, benchmark against KnetArrays
- Get rid of libknet8 calls, replace with Julia kernels, CUDA calls, Torch.jl etc.
- Get rid of @cu calls, replace them with CUDA calls
- addtoindex! ambiguity
- Warning: Hardware is unsupported by NNPACK so falling back to default NNlib
- stop using cudart, CUDA.jl uses cu interface, use its functions instead
- travis does not push new docs
- CuArrays.jl: DEPRECATED, use CUDA.jl instead!
- Performance: cudnn algorithm selection
- Performance: perceptron
- Performance: elementwise operations
- Display for CuArray within Tuples does not respect :limit=>true
- Performance: getindex(a, i::Array{Int})
- Performance: bias add
- On the use of @sync during benchmarking in the documentation
- Can we pleeeeeeeease make cu(x) eltype preserving?
- Deprecation warnings