apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-12-24 13:28:13 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Remove CUDA ERROR 9 of inputs of get_padding_offset kernel (#5440)

Browse Source 
Co-authored-by: K11OntheBoat <“ruianmaidanglao@163.com”>
This commit is contained in:
K11OntheBoat
2025-12-09 14:17:30 +08:00
committed by GitHub
parent 76649b45c1
commit 8d99bac532
8 changed files with 97 additions and 167 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
custom_ops/gpu_ops/cpp_extensions.cc
View File

@@ -388,8 +388,8 @@ void GetBlockShapeAndSplitKVBlock(
const int block_size);
std::vector<paddle::Tensor> GetPaddingOffset(const paddle::Tensor& input_ids,
const paddle::Tensor& token_num,
const paddle::Tensor& seq_len);
const paddle::Tensor& seq_len,
const int64_t token_num_cpu);
void SetValueByFlagsAndIdx(const paddle::Tensor& pre_ids_all,
const paddle::Tensor& input_ids,
@@ -725,9 +725,9 @@ std::vector<paddle::Tensor> SpeculateGetPaddingOffset(
const paddle::Tensor& input_ids,
const paddle::Tensor& draft_tokens,
const paddle::Tensor& cum_offsets,
const paddle::Tensor& token_num,
const paddle::Tensor& seq_len,
const paddle::Tensor& seq_lens_encoder);
const paddle::Tensor& seq_lens_encoder,
const int64_t token_num_cpu);
std::vector<paddle::Tensor> SpeculateGetSeqLensOutput(
const paddle::Tensor& seq_lens_this_time,