-
Notifications
You must be signed in to change notification settings - Fork 1.7k
[TRTLLM-5208][BREAKING CHANGE] chore: make pytorch LLM the default #5312
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.
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
Conversation
cedc6ea
to
ec3fd21
Compare
5ca2c7b
to
42c2cad
Compare
Since we have already branch out 0.21 branch, it is okay to land this PR onto the GH main directly. Thanks |
/bot run |
PR_Github #9377 [ run ] triggered by Bot |
PR_Github #9377 [ run ] completed with state |
/bot run --disable-fail-fast |
PR_Github #9394 [ run ] triggered by Bot |
c4ea41a
to
5ef186c
Compare
/bot run --disable-fail-fast |
PR_Github #9461 [ run ] triggered by Bot |
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
PR_Github #9461 [ run ] completed with state |
5ef186c
to
9595180
Compare
/bot run --disable-fail-fast |
PR_Github #9499 [ run ] triggered by Bot |
PR_Github #9499 [ run ] completed with state |
Signed-off-by: Superjomn <[email protected]>
Signed-off-by: Superjomn <[email protected]>
9595180
to
302d4b1
Compare
/bot run --disable-fail-fast |
PR_Github #9525 [ run ] triggered by Bot |
PR_Github #9525 [ run ] completed with state |
…VIDIA#5312) Signed-off-by: Superjomn <[email protected]>
…VIDIA#5312) Signed-off-by: Superjomn <[email protected]>
…VIDIA#5312) Signed-off-by: Superjomn <[email protected]>
…VIDIA#5312) Signed-off-by: Superjomn <[email protected]>
…VIDIA#5312) Signed-off-by: Superjomn <[email protected]>
…VIDIA#5312) Signed-off-by: Superjomn <[email protected]>
…VIDIA#5312) Signed-off-by: Superjomn <[email protected]>
…VIDIA#5312) Signed-off-by: Superjomn <[email protected]>
PR Description
This PR makes the PyT LLM the default with the following API breaking change which is confirmed with @laikhtewari :
from tensorrt_llm import LLM
from tensorrt_llm._tensorrt_engine import LLM
We introduce the need for an explicit code change to existing TRT backend users due to the arglists of both PyT LLM and TRT LLM diverging, and there is no seamless way to switch the backend without modifying the code.
The usage code, including tests and examples, has been minimally updated to keep the PR concise and focused.
There will be dedicated PRs to change the doc, examples accordingly later.