dylanyunlong/project_6
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

fix(build): std::optional -> c10::optional in ALL ilu/ files including ixformer.h

Browse Source
This commit is contained in:
claude
2026-08-11 07:40:42 +00:00
parent f944ef912b
commit c31a749143
11 changed files with 77 additions and 77 deletions
Download Patch File Download Diff File Expand all files Collapse all files

38
ex_engine/csrc/ilu/attention.cpp
View File

@@ -23,9 +23,9 @@ using namespace ixformer;
namespace xllm::kernel::ilu {
void reshape_paged_cache(torch::Tensor& key,
std::optional<torch::Tensor>& value,
c10::optional<torch::Tensor>& value,
torch::Tensor& key_cache,
std::optional<torch::Tensor>& value_cache,
c10::optional<torch::Tensor>& value_cache,
torch::Tensor& slot_mapping) {
auto value_ = value.value_or(torch::Tensor());
auto value_cache_ = value_cache.value_or(torch::Tensor());
@@ -47,16 +47,16 @@ void reshape_paged_cache(torch::Tensor& key,
void batch_prefill(torch::Tensor& query,
const torch::Tensor& key,
const std::optional<torch::Tensor>& value,
const c10::optional<torch::Tensor>& value,
torch::Tensor& output,
std::optional<torch::Tensor>& output_lse,
const std::optional<torch::Tensor>& q_cu_seq_lens,
const std::optional<torch::Tensor>& kv_cu_seq_lens,
const std::optional<torch::Tensor>& alibi_slope,
const std::optional<torch::Tensor>& attn_bias,
const std::optional<torch::Tensor>& q_quant_scale,
const std::optional<torch::Tensor>& k_quant_scale,
const std::optional<torch::Tensor>& v_quant_scale,
c10::optional<torch::Tensor>& output_lse,
const c10::optional<torch::Tensor>& q_cu_seq_lens,
const c10::optional<torch::Tensor>& kv_cu_seq_lens,
const c10::optional<torch::Tensor>& alibi_slope,
const c10::optional<torch::Tensor>& attn_bias,
const c10::optional<torch::Tensor>& q_quant_scale,
const c10::optional<torch::Tensor>& k_quant_scale,
const c10::optional<torch::Tensor>& v_quant_scale,
const torch::Tensor& block_tables,
int64_t max_query_len,
int64_t max_seq_len,
@@ -101,14 +101,14 @@ void batch_decode(torch::Tensor& query,
torch::Tensor& output,
const torch::Tensor& block_table,
const torch::Tensor& seq_lens,
const std::optional<torch::Tensor>& v_cache,
std::optional<torch::Tensor>& output_lse,
const std::optional<torch::Tensor>& q_quant_scale,
const std::optional<torch::Tensor>& k_cache_quant_scale,
const std::optional<torch::Tensor>& v_cache_quant_scale,
const std::optional<torch::Tensor>& out_quant_scale,
const std::optional<torch::Tensor>& alibi_slope,
const std::optional<torch::Tensor>& mask,
const c10::optional<torch::Tensor>& v_cache,
c10::optional<torch::Tensor>& output_lse,
const c10::optional<torch::Tensor>& q_quant_scale,
const c10::optional<torch::Tensor>& k_cache_quant_scale,
const c10::optional<torch::Tensor>& v_cache_quant_scale,
const c10::optional<torch::Tensor>& out_quant_scale,
const c10::optional<torch::Tensor>& alibi_slope,
const c10::optional<torch::Tensor>& mask,
const std::string& compute_dtype,
int64_t max_seq_len,
int64_t window_size_left,