kanshan/TensorRT-LLMs
1
0
Fork 0
mirror of https://github.com/NVIDIA/TensorRT-LLM.git synced 2026-02-05 02:31:33 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity
3aaed62cfc
TensorRT-LLMs/tensorrt_llm/_torch/speculative
History
Stefan Niebler 0cfd08745c
[TRTLLM-9735][feat] Add processed logprobs functionality to TorchSampler (#9675) 
Signed-off-by: Stefan Niebler <82932102+stnie@users.noreply.github.com>
Signed-off-by: Yuan Tong <13075180+tongyuantongyu@users.noreply.github.com>
Signed-off-by: Erin Ho <14718778+hchings@users.noreply.github.com>
Co-authored-by: Yuan Tong <13075180+tongyuantongyu@users.noreply.github.com>
Co-authored-by: Erin Ho <14718778+hchings@users.noreply.github.com>
2026-01-16 10:52:41 -08:00
..
__init__.py [None][feat] Implement sampling for MTP 1-model (#10019) 2025-12-31 13:48:34 -05:00
auto_heuristic.py [None][feat] Clean up ngram auto mode, add max_concurrency to configs (#6676) 2025-08-07 12:51:47 -04:00
drafter.py [TRTLLM-8136][feat] Dynamic draft length in spec decode (stage 1). (#8194) 2025-11-18 11:13:39 -05:00
drafting_loops.py [https://nvbugs/5772414][fix] Fix draft token tree depth=1 corner case (#10385) 2026-01-05 17:20:14 +01:00
eagle3.py [None][feat] Eagle: MLA Based Eagle (#9677) 2026-01-02 13:45:07 -05:00
interface.py [https://nvbugs/5781589][fix] Implement pp skip forward for all spec workers. (#10578) 2026-01-14 09:36:35 +08:00
model_drafter.py [TRTLLM-9962][feat] Some optimizations for two-model spec dec (#10208) 2025-12-28 12:52:04 +08:00
mtp.py [TRTLLM-9735][feat] Add processed logprobs functionality to TorchSampler (#9675) 2026-01-16 10:52:41 -08:00
ngram.py [TRTLLM-8160][feat] Add draft token tree runtime on CDL (#8586) 2025-11-25 09:40:55 -05:00
one_model_sampler.py [None][feat] Implement sampling on 1-model EAGLE3 (#9885) 2025-12-13 07:38:22 -08:00
save_hidden_state.py [TRTLLM-8160][feat] Add max_total_draft_tokens (#8366) 2025-10-21 11:11:04 -04:00
spec_tree_manager.py [TRTLLM-8160][feat] Add draft token tree runtime on CDL (#8586) 2025-11-25 09:40:55 -05:00
speculation_gate.py [TRTLLM-7412][feat] Turn off spec decode when the rolling average acceptance length drops below threshold. (#7283) 2025-10-13 15:51:14 -07:00
utils.py [TRTLLM-9416][feat] Skip DS-v3.2 indexer MQA and Top-K for short sequences. (#9524) 2025-12-15 12:42:25 +08:00