-
Notifications
You must be signed in to change notification settings - Fork 235
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
update gaudi3 baseline numbers for ci and py tests #1821
base: transformers_future
Are you sure you want to change the base?
Conversation
Co-authored-by: regisss <[email protected]>
…ion (#1799) Co-authored-by: regisss <[email protected]>
...otherwise the pytest collect will fail and no tests are executed. Use skipif decorator, instead. Signed-off-by: U. Artie Eoff <[email protected]>
...otherwise the pytest collect can fail and no tests are executed. Install from test body, instead. Signed-off-by: U. Artie Eoff <[email protected]>
Signed-off-by: U. Artie Eoff <[email protected]>
Signed-off-by: U. Artie Eoff <[email protected]>
...and rename "gaudi" to gaudi1" so we can use the new baseline fixture. Signed-off-by: U. Artie Eoff <[email protected]>
We want to run the same tests for Gaudi2 on Gaudi3. Hence, Gaudi1 is the exception case... so reverse test definition conditional logic as needed. Signed-off-by: U. Artie Eoff <[email protected]>
Co-authored-by: Yaser Afshar <[email protected]>
Start with the same references on gaudi3 as gaudi2. Then, we can incrementally update them as needed. Signed-off-by: U. Artie Eoff <[email protected]>
Signed-off-by: U. Artie Eoff <[email protected]>
}, | ||
"gaudi3": { | ||
"output": "DeepSpeed is a machine learning framework that enables you to train large models on a single GPU. It is a framework that is used to train large models on a single GPU.\n\nThe main idea is to use a large amount of memory to fit the model on a single GPU.\n\nThe main idea is to use a large amount of memory to fit the model on a single GPU.\n\nThe main idea is to use a large amount of memory to fit the model on a single GPU.\n\nDeepSpeed is a framework that allows you", | ||
"throughput": 46.04685368495098 |
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.
seems this duplication is an issue. @regisss can you also check if this is an issue with greedy search?
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.
This can happen with basic greedy search, it's not really an issue. We would need to tweak a few generation parameters to improve it. I don't have the bandwidth to do it this week, but I can look into it next week.
update gaudi3 baseline numbers for ci and py tests