EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

Add typo checker in pre-commit (#6179)

Browse Source 
Co-authored-by: Brayden Zhong <b8zhong@uwaterloo.ca>
This commit is contained in:
applesaucethebun
2025-05-11 00:55:00 -04:00
committed by GitHub
parent de167cf5fa
commit 2ce8793519
99 changed files with 154 additions and 144 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
sgl-kernel/csrc/cpu/common.h
View File

@@ -49,7 +49,7 @@ namespace {
#define CHECK_CONTIGUOUS(x) TORCH_CHECK(x.is_contiguous(), #x " must be contiguous")
#define CHECK_LAST_DIM_CONTIGUOUS(x) \
TORCH_CHECK(x.strides()[x.strides().size() - 1] == 1, #x "must be contiguous at last dimention")
TORCH_CHECK(x.strides()[x.strides().size() - 1] == 1, #x "must be contiguous at last dimension")
#define CHECK_INPUT(x) \
CHECK_CPU(x); \

4
sgl-kernel/csrc/cpu/decode.cpp
View File

@@ -718,7 +718,7 @@ void decode_attention_kernel_impl(
m_prime = m_i;
// caculate V' <- s_delta @ V + V' * m_delta
// calculate V' <- s_delta @ V + V' * m_delta
index_gemm_kernel_nn<scalar_t, index_t>(
/* A */ s_delta,
/* B */ v_buffer + head_id * v_strideH,
@@ -925,7 +925,7 @@ void decode_attention_grouped_kernel_impl(
m_prime[h] = m_i;
}
// caculate V' <- s_delta @ V + V' * m_delta
// calculate V' <- s_delta @ V + V' * m_delta
index_gemm_kernel_nn<scalar_t, index_t>(
/* A */ s_delta,
/* B */ v_buffer + head_kv_id * v_strideH,

4
sgl-kernel/csrc/cpu/extend.cpp
View File

@@ -323,7 +323,7 @@ void extend_attention_kernel_impl(
/* ld_src */ v_strideN,
/* ld_dst */ head_size_v);
// caculate V' <- s_delta @ V + V'
// calculate V' <- s_delta @ V + V'
at::native::cpublas::brgemm(
/* M */ m_size,
/* N */ head_size_v,
@@ -434,7 +434,7 @@ void extend_attention_kernel_impl(
/* ld_src */ ve_strideN,
/* ld_dst */ head_size_v);
// caculate V' <- s_delta @ V + V'
// calculate V' <- s_delta @ V + V'
at::native::cpublas::brgemm(
/* M */ m_size,
/* N */ head_size_v,

2
sgl-kernel/csrc/cpu/gemm.h
View File

@@ -79,7 +79,7 @@ void fused_experts_int8_kernel_impl(
int64_t topk,
int64_t num_tokens_post_pad);
// shared expert implememntation for int8 w8a8
// shared expert implementation for int8 w8a8
template <typename scalar_t>
void shared_expert_int8_kernel_impl(
scalar_t* __restrict__ output,

2
sgl-kernel/csrc/cpu/gemm_int8.cpp
View File

@@ -51,7 +51,7 @@ struct tinygemm_kernel_nn<at::BFloat16, has_bias, BLOCK_M, BLOCK_N> {
__m512 vd0;
__m512 vd1[COLS];
// oops! 4x4 spills but luckly we use 4x2
// oops! 4x4 spills but we use 4x2
__m512 vbias[COLS];
// [NOTE]: s8s8 igemm compensation in avx512-vnni