Skip to content

fix mtp.py typo #9307

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

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
+3 −3
Open

fix mtp.py typo #9307

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
6 changes: 3 additions & 3 deletions tensorrt_llm/_torch/speculative/mtp.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -406,20 +406,20 @@ def forward(
- KV cache: (ABCD) + EFGH (H's KV cache is invalid)
- hidden states: H_E, H_F, H_G, H_H (H_H is invalid)
Draft model:
MPT1:
MTP1:
# For generation request, `mtp_num_modules` of tokens will be used as input.
- input tokens: FGX
- input hidden states: H_E, H_F, H_G
- KV cache: (BCDE) + FGX
- output hidden states: h_F, h_G, h_X
- output next draft token: N
MPT2:
MTP2:
- input tokens: GXN
- input hidden states: H_F, H_G, h_X
- KV cache: (CDEF) + GXN
- output hidden states: h_G, h_X, h_N
- output next draft token: O
MPT3:
MTP3:
- input tokens: XNO
- input hidden states: H_G, H_X, h_N
- KV cache: (DEFG) + XNO
Expand Down