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

del batch id per token (#3963)

Browse Source 
* Update decoder_write_cache_with_rope_kernel.cu

del batch_id_per_token

* Update decoder_write_cache_with_rope_impl.cuh

* Update test_append_attention.py

* Update test_append_attention.py
This commit is contained in:
lzy
2025-09-08 21:58:34 +08:00
committed by GitHub
parent 08b3153661
commit f12159b630
3 changed files with 7 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
custom_ops/gpu_ops/append_attn/decoder_write_cache_with_rope_impl.cuh
View File

@@ -684,7 +684,6 @@ __global__ void append_decode_cache_int8_rope_qk_norm_kernel(
// block_size, head_size // 2]
T* __restrict__ qkv_out,
const int* __restrict__ block_tables, // [bsz, max_blocks_per_seq]
const int* __restrict__ batch_id_per_token, // [num_tokens]
const int* __restrict__ cu_seqlens_q,
const int* __restrict__ seq_lens, // [bsz]
const int* __restrict__ seq_lens_encoder, // [bsz]

2
custom_ops/gpu_ops/append_attn/decoder_write_cache_with_rope_kernel.cu
View File

@@ -565,7 +565,6 @@ void DecoderWriteCacheWithRoPEKernel(
value_cache_out->data<uint8_t>(),
reinterpret_cast<DataType_*>(qkv_out->data<T>()),
block_tables.data<int>(),
batch_id_per_token.data<int>(),
cu_seqlens_q.data<int>(),
seq_lens.data<int>(),
seq_lens_encoder.data<int>(),
@@ -729,7 +728,6 @@ void DecoderWriteCacheWithRoPEKernel(
value_cache_out->data<uint8_t>(),
reinterpret_cast<DataType_*>(qkv_out->data<T>()),
block_tables.data<int>(),
batch_id_per_token.data<int>(),
cu_seqlens_q.data<int>(),
seq_lens.data<int>(),
seq_lens_encoder.data<int>(),