AlpinDale 0e6c400b13 feat: re-add GGUF (#600) hai 6 meses
..
all_reduce 156f577f79 feat: switch from `PYBIND11_MODULE` to `TORCH_LIBRARY` (#569) hai 7 meses
attention dc1b59df9c fix: compiler warnings for _C and _moe hai 6 meses
backup f8dfac6372 chore: attention refactor and upstream sync apr01 (#365) hai 11 meses
core 141672a0d4 kernels: disambiguate quantized types via a new ScalarType hai 6 meses
cpu 141672a0d4 kernels: disambiguate quantized types via a new ScalarType hai 6 meses
hadamard 5d288aa76c feat: add fast hadamard transformation kernels (#232) hai 1 ano
mamba dc1b59df9c fix: compiler warnings for _C and _moe hai 6 meses
moe 141672a0d4 kernels: disambiguate quantized types via a new ScalarType hai 6 meses
prepare_inputs dd18c5042c move prepare_inputs to the GPU (#596) hai 7 meses
punica 848731f527 chore: add punica sizes for mistral nemo hai 6 meses
quantization 0e6c400b13 feat: re-add GGUF (#600) hai 6 meses
activation_kernels.cu 5fbd93797d fix: beta value in gelu_tanh kernel being divided by 0.5 hai 7 meses
cache.h 32bdbd1ee4 chore: add fp8 support to `reshape_and_cache_flash` hai 7 meses
cache_kernels.cu 32bdbd1ee4 chore: add fp8 support to `reshape_and_cache_flash` hai 7 meses
cuda_compat.h 00acf371f9 rocm: fused topk softmax hai 7 meses
cuda_utils.h 156f577f79 feat: switch from `PYBIND11_MODULE` to `TORCH_LIBRARY` (#569) hai 7 meses
cuda_utils_kernels.cu 156f577f79 feat: switch from `PYBIND11_MODULE` to `TORCH_LIBRARY` (#569) hai 7 meses
dispatch_utils.h dc1b59df9c fix: compiler warnings for _C and _moe hai 6 meses
layernorm_kernels.cu dc1b59df9c fix: compiler warnings for _C and _moe hai 6 meses
ops.h dd18c5042c move prepare_inputs to the GPU (#596) hai 7 meses
pos_encoding_kernels.cu 156f577f79 feat: switch from `PYBIND11_MODULE` to `TORCH_LIBRARY` (#569) hai 7 meses
reduction.cuh aba03b4756 feat: dynamic per-token activation quantization hai 7 meses
torch_bindings.cpp 0e6c400b13 feat: re-add GGUF (#600) hai 6 meses