biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Model][Spec Decode] Nemotron-H MTP and Mamba Speculative Decoding Support (#33726)

Browse Source 
Signed-off-by: Shahar Mor <smor@nvidia.com>
Signed-off-by: Benjamin Chislett <bchislett@nvidia.com>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Co-authored-by: Shahar Mor <smor@nvidia.com>
Co-authored-by: Roi Koren <roik@nvidia.com>
Co-authored-by: Lucas Wilkinson <lwilkins@redhat.com>
This commit is contained in:
Benjamin Chislett
2026-02-24 12:49:56 -05:00
committed by GitHub
parent a9e15e040d
commit f5972a872f
19 changed files with 799 additions and 157 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
vllm/model_executor/models/registry.py
View File

@@ -522,6 +522,7 @@ _SPECULATIVE_DECODING_MODELS = {
"DeepSeekMTPModel": ("deepseek_mtp", "DeepSeekMTP"),
"ErnieMTPModel": ("ernie_mtp", "ErnieMTP"),
"ExaoneMoeMTP": ("exaone_moe_mtp", "ExaoneMoeMTP"),
"NemotronHMTPModel": ("nemotron_h_mtp", "NemotronHMTP"),
"LongCatFlashMTPModel": ("longcat_flash_mtp", "LongCatFlashMTP"),
"Glm4MoeMTPModel": ("glm4_moe_mtp", "Glm4MoeMTP"),
"Glm4MoeLiteMTPModel": ("glm4_moe_lite_mtp", "Glm4MoeLiteMTP"),