Q4_2 quantization with rmse-optimized scale and quants #1062
+87
−3
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
For quantize-stats we get q4_2: rmse 0.00159301, maxerr 0.17480469, 95pct<0.0030, median<0.0012 For 7B perplexity with BLAS enabled we get 6.2038 after 655 chunks. Quantization is slow (~90 seconds on my Mac for 7B) as not multi-threaded as in PR #896.
ggerganov
approved these changes
Apr 19, 2023
Not sure why this makes them fail
sw
reviewed
Apr 19, 2023
sw
reviewed
Apr 19, 2023
Green-Sky
reviewed
Apr 19, 2023
| @@ -1123,12 +1124,94 @@ static void quantize_row_q4_2_reference(const float * restrict x, block_q4_2 * r | |||
| } | |||
| } | |||
|
|
|||
| static inline int nearest_int(float fval) { | |||
There was a problem hiding this comment.
this inline does not do anything here. the static is all you need.
There was a problem hiding this comment.
hm actually, after looking at cppref, i am not sure that C and C++ are the same here.
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Outdated suggestions cannot be applied.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.
For quantize-stats we get
q4_2: rmse 0.00159301, maxerr 0.17480469, 95pct<0.0030, median<0.0012
For 7B perplexity with BLAS enabled we get 6.2038 after 655 chunks.
Quantization is slow (~90 seconds on my Mac for 7B) as not multi-threaded as in PR #896.