Remove is_gpt_fast flag #172
Merged
Conversation
This file contains 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
facebook-github-bot
added
the
CLA Signed
Summary: It was added before to merge the code for 8da4w and int4 weight only quant, but later we just duplicated the quantizer code, so we can safely remove this now. in the future we'll refactor everything to use tensor subclass. Test Plan: tested locally to make sure `test_8da4w_quantizer_eval` still works Reviewers: Subscribers: Tasks: Tags:
cpuhrsch
approved these changes
Apr 24, 2024
msaroufim
reviewed
Apr 24, 2024
| @@ -268,57 +268,6 @@ def test_8da4w_quantizer_eval(self): | |||
| f"accuracy regressed from 8.23 to {result['results']['wikitext']['word_perplexity,none']}" | |||
| ) | |||
|
|
|||
| @unittest.skip("skipping until we get checkpoints for gpt-fast") | |||
| def test_gptq_quantizer_gpt_fast(self): | |||
There was a problem hiding this comment.
is this test not useful to keep around or some new version of it? Or more generally can we no longer use gpt-fast and ao together?
There was a problem hiding this comment.
oh next test tests gpt fast code path: test_gptq_quantizer_int4wo, this was initially added because we were trying to merge the gpt fast code path and 8da4w code path together in the same quantizer code path and use a flag to distinguish them, but now we just duplicated the quantizer code
so since we removed the gpt-fast code path in Int8DynActInt4WeightGPTQQuantizer, we no longer need to test this
msaroufim
approved these changes
Apr 24, 2024
dbyoung18
pushed a commit
to dbyoung18/ao
that referenced
this pull request
Jul 31, 2024
Summary: It was added before to merge the code for 8da4w and int4 weight only quant, but later we just duplicated the quantizer code, so we can safely remove this now. in the future we'll refactor everything to use tensor subclass. Test Plan: tested locally to make sure `test_8da4w_quantizer_eval` still works Reviewers: Subscribers: Tasks: Tags:
yanbing-j
pushed a commit
to yanbing-j/ao
that referenced
this pull request
Dec 9, 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.
Summary:
It was added before to merge the code for 8da4w and int4 weight only quant, but later we just duplicated the quantizer code, so we can safely remove this now.
in the future we'll refactor everything to use tensor subclass.
Test Plan:
tested locally to make sure
test_8da4w_quantizer_evalstill worksReviewers:
Subscribers:
Tasks:
Tags: