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

[BugFix] Rename attention params of deepseekv3 (#2939)

Browse Source 
Co-authored-by: K11OntheBoat <“ruianmaidanglao@163.com”>
This commit is contained in:
K11OntheBoat
2025-07-22 14:01:30 +08:00
committed by GitHub
parent 56102e91e1
commit 8020927f50
7 changed files with 43 additions and 44 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
custom_ops/gpu_ops/append_attn/mla_cache_kernel.cu
View File

@@ -91,7 +91,7 @@ std::vector<paddle::Tensor> PrefillMLAWriteCacheKernel(
meta_data.max_blocks_per_seq = block_tables.dims()[1];
meta_data.block_size = kv_cache_dims[2];
meta_data.batch_size = cu_seqlens_q.dims()[0];
meta_data.batch_size = seq_lens_decoder.dims()[0];
switch (kv_pe.dtype()) {
case paddle::DataType::BFLOAT16: {
return PrefillMLAWriteCache<paddle::DataType::BFLOAT16>(meta_data,
@@ -224,7 +224,7 @@ std::vector<paddle::Tensor> DecodeMLAWriteCacheKernel(
meta_data.max_blocks_per_seq = block_tables.dims()[1];
meta_data.block_size = kv_cache_dims[2];
meta_data.batch_size = cu_seqlens_q.dims()[0];
meta_data.batch_size = seq_lens_encoder.dims()[0];
switch (kv_pe.dtype()) {
case paddle::DataType::BFLOAT16: {
return DecodeMLAWriteCache<paddle::DataType::BFLOAT16>(meta_data,