-
Notifications
You must be signed in to change notification settings - Fork 211
[CPP Graph] Baichuan & Baichuan2 Enabling #376
Conversation
intel_extension_for_transformers/llm/runtime/graph/application/main_run.cpp
Outdated
Show resolved
Hide resolved
please update KERNEL_DEBUG settings as zhenwei issued during the meeting, and prepare readme |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM except some comments
intel_extension_for_transformers/llm/runtime/graph/models/baichuan/baichuan.h
Outdated
Show resolved
Hide resolved
intel_extension_for_transformers/llm/runtime/graph/models/baichuan/baichuan_utils.cpp
Outdated
Show resolved
Hide resolved
intel_extension_for_transformers/llm/runtime/graph/models/baichuan/baichuan_utils.cpp
Show resolved
Hide resolved
better to finish this in this PR |
I've removed this feature in this PR, because printing data info using a unified API is not something that can be solved with a few simple lines of code, which means it requires more efforts. |
intel_extension_for_transformers/llm/runtime/graph/models/baichuan/baichuan_utils.cpp
Outdated
Show resolved
Hide resolved
Do we enable FFN fusion / MHA fusion in this pr? |
No. The MHA & FFN fusion to the chatglm & baichuan will be applied in the next PR. I plan to add models first this week. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
Type of Change
Feature add and Issues Fix
Description
Add extension-test
Expected Behavior & Potential Risk
N/A.
How has this PR been tested?
Manual and CI tests.
Dependency Change?
N/A.