biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[CI] change spell checker from codespell to typos (#18711)

Browse Source 
Signed-off-by: Andy Xie <andy.xning@gmail.com>
This commit is contained in:
Ning Xie
2025-06-12 10:57:10 +08:00
committed by GitHub
parent 42f52cc95b
commit 2f1c19b245
57 changed files with 335 additions and 163 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
csrc/moe/moe_permute_unpermute_op.cu
View File

@@ -12,7 +12,7 @@ void moe_permute(
const torch::Tensor& input, // [n_token, hidden]
const torch::Tensor& topk_weights, //[n_token, topk]
torch::Tensor& topk_ids, // [n_token, topk]
const torch::Tensor& token_expert_indicies, // [n_token, topk]
const torch::Tensor& token_expert_indices, // [n_token, topk]
const std::optional<torch::Tensor>& expert_map, // [n_expert]
int64_t n_expert, int64_t n_local_expert, int64_t topk,
const std::optional<int64_t>& align_block_size,
@@ -27,15 +27,15 @@ void moe_permute(
"expert_first_token_offset must be int64");
TORCH_CHECK(topk_ids.scalar_type() == at::ScalarType::Int,
"topk_ids must be int32");
TORCH_CHECK(token_expert_indicies.scalar_type() == at::ScalarType::Int,
"token_expert_indicies must be int32");
TORCH_CHECK(token_expert_indices.scalar_type() == at::ScalarType::Int,
"token_expert_indices must be int32");
TORCH_CHECK(src_row_id2dst_row_id_map.scalar_type() == at::ScalarType::Int,
"src_row_id2dst_row_id_map must be int32");
TORCH_CHECK(expert_first_token_offset.size(0) == n_local_expert + 1,
"expert_first_token_offset shape != n_local_expert+1")
TORCH_CHECK(
src_row_id2dst_row_id_map.sizes() == token_expert_indicies.sizes(),
"token_expert_indicies shape must be same as src_row_id2dst_row_id_map");
src_row_id2dst_row_id_map.sizes() == token_expert_indices.sizes(),
"token_expert_indices shape must be same as src_row_id2dst_row_id_map");
auto n_token = input.sizes()[0];
auto n_hidden = input.sizes()[1];
auto align_block_size_value =
@@ -71,7 +71,7 @@ void moe_permute(
expert_map_ptr, n_expert, stream);
}
// expert sort topk expert id and scan expert id get expert_first_token_offset
sortAndScanExpert(get_ptr<int>(topk_ids), get_ptr<int>(token_expert_indicies),
sortAndScanExpert(get_ptr<int>(topk_ids), get_ptr<int>(token_expert_indices),
get_ptr<int>(permuted_experts_id),
get_ptr<int>(dst_row_id2src_row_id_map),
get_ptr<int64_t>(expert_first_token_offset), n_token,
@@ -190,7 +190,7 @@ void shuffle_rows(const torch::Tensor& input_tensor,
void moe_permute(const torch::Tensor& input, const torch::Tensor& topk_weights,
torch::Tensor& topk_ids,
const torch::Tensor& token_expert_indicies,
const torch::Tensor& token_expert_indices,
const std::optional<torch::Tensor>& expert_map,
int64_t n_expert, int64_t n_local_expert, int64_t topk,
const std::optional<int64_t>& align_block_size,
@@ -203,7 +203,7 @@ void moe_permute(const torch::Tensor& input, const torch::Tensor& topk_weights,
void moe_unpermute(const torch::Tensor& input,
const torch::Tensor& topk_weights, torch::Tensor& topk_ids,
const torch::Tensor& token_expert_indicies,
const torch::Tensor& token_expert_indices,
const std::optional<torch::Tensor>& expert_map,
int64_t n_expert, int64_t n_local_expert, int64_t topk,
const std::optional<int64_t>& align_block_size,

6
csrc/moe/topk_softmax_kernels.cu
View File

@@ -425,7 +425,7 @@ void topkGatingSoftmaxLauncherHelper(const float* input, const bool* finished, f
#define LAUNCH_SOFTMAX(NUM_EXPERTS, WARPS_PER_TB) \
topkGatingSoftmaxLauncherHelper<NUM_EXPERTS, WARPS_PER_TB>( \
gating_output, nullptr, topk_weights, topk_indicies, \
gating_output, nullptr, topk_weights, topk_indices, \
token_expert_indices, num_tokens, topk, 0, num_experts, \
stream);
@@ -433,7 +433,7 @@ template <typename IndType>
void topkGatingSoftmaxKernelLauncher(
const float* gating_output,
float* topk_weights,
IndType* topk_indicies,
IndType* topk_indices,
int* token_expert_indices,
float* softmax_workspace,
const int num_tokens,
@@ -476,7 +476,7 @@ void topkGatingSoftmaxKernelLauncher(
moeSoftmax<TPB><<<num_tokens, TPB, 0, stream>>>(
gating_output, nullptr, softmax_workspace, num_experts);
moeTopK<TPB><<<num_tokens, TPB, 0, stream>>>(
softmax_workspace, nullptr, topk_weights, topk_indicies, token_expert_indices,
softmax_workspace, nullptr, topk_weights, topk_indices, token_expert_indices,
num_experts, topk, 0, num_experts);
}
}

2
csrc/moe/torch_bindings.cpp
View File

@@ -66,7 +66,7 @@ TORCH_LIBRARY_EXPAND(TORCH_EXTENSION_NAME, m) {
m.def(
"moe_permute(Tensor input, Tensor topk_weight, Tensor! topk_ids,"
"Tensor token_expert_indicies, Tensor? expert_map, int n_expert,"
"Tensor token_expert_indices, Tensor? expert_map, int n_expert,"
"int n_local_expert,"
"int topk, int? align_block_size,Tensor! permuted_input, Tensor! "
"expert_first_token_offset, Tensor! src_row_id2dst_row_id_map, Tensor! "