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

support speculate_limit_thinking_content_length_v2 (#4428)
Some checks failed
CE Compile Job / ce_job_pre_check (push) Has been cancelled
Details
CE Compile Job / print_ce_job_pre_check_outputs (push) Has been cancelled
Details
CE Compile Job / FD-Clone-Linux (push) Has been cancelled
Details
CE Compile Job / Show Code Archive Output (push) Has been cancelled
Details
CE Compile Job / BUILD_SM8090 (push) Has been cancelled
Details
CE Compile Job / BUILD_SM8689 (push) Has been cancelled
Details
CE Compile Job / CE_UPLOAD (push) Has been cancelled
Details

Browse Source 
* support speculate_limit_thinking_content_length_v2

* fix

* fix import
This commit is contained in:
Yuanle Liu
2025-10-16 13:23:16 +08:00
committed by GitHub
parent 0982dfb705
commit 83f97d1196
4 changed files with 193 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
custom_ops/gpu_ops/cpp_extensions.cc
View File

@@ -678,6 +678,16 @@ void SpeculateVerify(
const paddle::Tensor &actual_draft_token_nums, const paddle::Tensor &topp,
int max_seq_len, int verify_window, bool enable_topp, bool benchmark_mode, bool accept_all_drafts);
void SpeculateLimitThinkingContentLengthV2(
const paddle::Tensor& next_tokens,
const paddle::Tensor& max_think_lens,
const paddle::Tensor& step_idx,
const paddle::Tensor& limit_think_status,
const paddle::Tensor& accept_num,
const paddle::Tensor& seq_lens_decoder,
const int64_t think_end_id,
const int64_t line_break_id);
void SpeculateUpdate(const paddle::Tensor &seq_lens_encoder,
const paddle::Tensor &seq_lens_decoder,
const paddle::Tensor &not_need_stop,
@@ -1245,6 +1255,8 @@ PYBIND11_MODULE(fastdeploy_ops, m) {
m.def("speculate_verify",&SpeculateVerify, "speculate_verify function");
m.def("speculate_limit_thinking_content_length_v2",&SpeculateLimitThinkingContentLengthV2, "speculate limit thinking content length function");
m.def("speculate_update",&SpeculateUpdate, "Speculate Update Kernel");
m.def("speculate_set_value_by_flags_and_idx",&SpeculateSetValueByFlagsAndIdx, "speculate_set_value_by_flags_and_idx function");