-
https://www.youtube.com/watch?v=JQYP2l8x6oQ
-
Added KernelBot and openzl
-
Added OLMo
-
Added nvshmem.
-
Deleted RDMA-Core.
-
Added kraken.
-
Added nccl back.
-
Updated contributors* fields
-
Finished categories for later added repos.
-
Removed projects that may not have been a focus.
-
flash-linear-attention
-
triton-windows
-
cutile and tilus
-
gemlite
-
Triton-distributed
-
triton-runner and ort/onnx
-
Removed 3 repos that weren't as relevant
-
nccl, RaBitQ, and triSYCL
-
OpenCL-SDK and ZLUDA. Replaced neuronx-distributed-inference with aws-neuron-sdk
-
tritonparse and TensorRT
-
server and ccache
-
metaflow and ome
-
oneDPL and SYCL-Docs
-
dstack and tflite-micro
-
Triton-distributed and pocl
-
numba and AdaptiveCpp
-
pybind, cuda-python and torchdynamo
-
nvcc4jupyter and CU2CL
-
lightning-thunder and ondemand
-
hip and openevolve
-
nixl and rdma-core
-
burn and DeepSpeed
-
peft and ROCm
-
aiter, LMCache, Mooncake, torchtitan and ao
-
Spark, executorch and trl
-
Revert to rocm systems and ray commit
-
Weights and Bias merge conflict
-
Retention rate
-
rocm systems and ray
-
transformers and hhvm
-
wandb
-
Sorted by unique contributors in 2025
-
Fixed contributors to be int instead of string
-
Added unique contributors count
-
Cleanup + truss and unsloth
-
warp and verl
-
MCP and Modular
