CanReader

CanReader/FastNN

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

26 good first / help-wanted issues · Rust · last activity Oct 1, 2026

0 stars 0 forks 0 watchers Rust MIT License
autograd cuda deep-learning deep-learning-library gpu hacktoberfest high-performance machine-learning machine-learning-library neural-network neural-networks rust rust-library tensor transformer
View on GitHub
26 Open Issues Need Help Last updated: Oct 1, 2026

Open Issues Need Help

View All on GitHub
variance() returns NaN for an empty tensor 20 days ago
bug good first issue P3 area:tensor
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
Embedding::forward and Batch::labels turn bad ids into index 0 20 days ago
bug good first issue P3 area:nn area:data
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
Linear::no_bias still draws a bias from the RNG and throws it away 20 days ago
bug good first issue P3 area:nn
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
fastnn_cuda_sum launches an invalid grid when n == 0 25 days ago
bug good first issue P3 area:cuda
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
Pooling has no CUDA path, every pool round trips through the host 25 days ago
enhancement help wanted P2 area:cuda
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
Reduction backward passes force a device sync via grad.item() 25 days ago
enhancement help wanted P2 area:autograd
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
sum_axis and min cudaMalloc/cudaFree on every call, which syncs the device 25 days ago
enhancement help wanted P2 area:cuda
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
Dropout builds its mask on the host and uploads it every forward 25 days ago
enhancement help wanted P2 area:nn
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
PositionalEncoding re-uploads its whole table on every forward 25 days ago
enhancement help wanted P2 area:nn
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
CPU broadcast ops are single threaded and clone both operands 25 days ago
enhancement help wanted P2 area:tensor
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
The CUDA block cache is unbounded across size classes 25 days ago
enhancement help wanted P2 area:cuda
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
build.rs picks CUDA paths from the host target, not the build target 25 days ago
bug help wanted good first issue P3 area:build
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
A partial MNIST download gets cached forever 25 days ago
bug help wanted good first issue P1 area:data
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
Checkpoint writes are not atomic, a crash mid save destroys the previous one 25 days ago
bug help wanted good first issue P0 area:serialize
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
Centered RMSprop can take the square root of a negative number 25 days ago
bug help wanted good first issue P1 area:optim
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
PARALLEL_THRESHOLD ignores how expensive the op is 25 days ago
enhancement help wanted P2 area:tensor
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
Checkpoints are written and read 4 bytes at a time 25 days ago
enhancement help wanted P2 area:serialize
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
safetensors save holds the whole file in memory 25 days ago
enhancement help wanted P2 area:serialize
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
Document what stays alive in memory during backward() 25 days ago
documentation good first issue area:autograd
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
safetensors load is O(n^2) in the number of tensors 25 days ago
enhancement good first issue P3 area:serialize
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
Linear::new(0, n) panics deep inside rand with an unrelated message 25 days ago
bug good first issue P3 area:tensor
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
StepDecay and OneCycle accept values that divide by zero or underflow about 1 month ago
bug good first issue P3 area:optim
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
Zero length sequences panic with a message about cat about 1 month ago
bug good first issue P3 area:nn
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
sum_axis_keep and mean_axis_keep add an extra rank on 1-D input about 1 month ago
bug good first issue P1 area:tensor
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
permute() accepts an order that is not a permutation about 1 month ago
bug good first issue P0 area:tensor
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer
expand() does not validate the target shape about 1 month ago
bug good first issue P0 area:tensor
CanReader/FastNN
0

A high-performance neural network library built with Rust and CUDA for blazing-fast deep learning

Rust
#autograd#cuda#deep-learning#deep-learning-library#gpu#hacktoberfest#high-performance#machine-learning#machine-learning-library#neural-network#neural-networks#rust#rust-library#tensor#transformer