Skip to content

[https://nvbugs/6422334][fix] In PyExecutor._prepare_draft_requests, gate the py_last_draft_tokens snapshot… #16121

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
zhaoyangwang-nvidia merged 1 commit into from
Jul 14, 2026
Merged

[https://nvbugs/6422334][fix] In PyExecutor._prepare_draft_requests, gate the py_last_draft_tokens snapshot… #16121

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 3 additions & 1 deletion tensorrt_llm/_torch/pyexecutor/py_executor.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3966,7 +3966,9 @@ def _prepare_draft_requests(self):
LlmRequestState.DISAGG_GENERATION_INIT):
continue

req.py_last_draft_tokens = req.py_draft_tokens
# Skip DISAGG_GENERATION_INIT: snapshotting the dummy py_draft_tokens leaks stale state into the first real draft.
if req.state == LlmRequestState.GENERATION_IN_PROGRESS:
req.py_last_draft_tokens = req.py_draft_tokens

if self.max_total_draft_tokens > 0 and self.use_spec_decode and not req.py_disable_speculative_decoding:
req.py_draft_tokens = [0] * self.max_total_draft_tokens
Expand Down
2 changes: 0 additions & 2 deletions tests/integration/test_lists/waives.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -270,8 +270,6 @@ full:H100/test_e2e.py::test_ptp_quickstart_advanced_deepseek_multi_nodes[DeepSee
full:H100_PCIe/unittest/llmapi/test_llm_pytorch.py::test_llama_7b_multi_lora_evict_and_reload_lora_gpu_cache SKIP (https://nvbugs/5682551)
full:H20/accuracy/test_disaggregated_serving.py::TestDeepSeekV3Lite::test_auto_dtype[mtp_nextn=2-overlap_scheduler=False] SKIP (https://nvbugs/6345827)
full:H20/accuracy/test_disaggregated_serving.py::TestDeepSeekV3Lite::test_auto_dtype[mtp_nextn=2-overlap_scheduler=True] SKIP (https://nvbugs/6345827)
full:H20/accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_guided_decoding_with_eagle3[llguidance-eagle3_one_model=False] SKIP (https://nvbugs/6422334)
full:H20/accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_guided_decoding_with_eagle3[xgrammar-eagle3_one_model=False] SKIP (https://nvbugs/6422334)
full:H20/accuracy/test_llm_api_autodeploy.py::TestModelRegistryAccuracy::test_autodeploy_from_registry[nvidia_Llama-3.1-8B-Instruct-FP8-True] SKIP (https://nvbugs/6422351)
full:H20/accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_bfloat16[mtp_nextn=2-attention_dp=True-cuda_graph=True-overlap_scheduler=True-torch_compile=True-enable_chunked_prefill=True-v2_kv_cache=False] SKIP (https://nvbugs/6422343)
full:H20/accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_bfloat16[mtp_nextn=2-attention_dp=True-cuda_graph=True-overlap_scheduler=True-torch_compile=True-enable_chunked_prefill=True-v2_kv_cache=True] SKIP (https://nvbugs/6422343)
Expand Down
Loading