[Dataset]: Iterate through benchmark dataset once #48
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
@philschmid Hey there. Could you install the version of the branch |
markurtz
requested changes
Sep 11, 2024
There was a problem hiding this comment.
Looks good, minor NIT on removing typing overrides and we should add in a test case for the main pathway for this.
Two additional points on testing:
- For automation, we should add in a quick test to ensure for the sweep case that we looped the dataset correctly and on start of a new benchmark it is the first item from the dataset -- can be merged in with the previous main tests
- For manual, please validate locally that calling with a hugging face dataset as well as a simple local csv file (can download the csv from the files for the linked dataset) will work properly
- Running a constant rate will execute the exact number of requests as in the dataset along with the first and last prompts sent matching the first and last in the dataset
- Running a sweep with a smaller (10 prompt) csv file will properly cycle and each benchmark run starts with the same beginning prompt and ends with the end prompt
parfeniukink
force-pushed
the
parfeniukink/issues/44
branch
from
9c077ab to
51ceaa8
Compare
added 7 commits
September 16, 2024 10:43
* Main CLI parameters are updated
* `MaxRequestsType(ParamType)` is used as a custom `click` param type
* Request generators are updated
* Unit testes are added
* Smoke tests are added
parfeniukink
force-pushed
the
parfeniukink/issues/44
branch
from
a02561f to
656804f
Compare
|
Hey Guys, any idea on when we could get this merged? |
markurtz
approved these changes
Oct 8, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
issue link: #44
Summary:
--max-requestsCLI parameter now supports thedatasetvalue.MaxRequestsType(ParamType)is used as a customclickparam type to validate input dataRequestGenerator(ABC)now includes the abstract__len__that corresponds to the length of the dataset if it is supported.guidellm/utils/text.pyis fixed. There is no reason to check if the file start with the'http'string. Basically this is a Bug since you never read a text file properly because of oldif/elsecondition.