Add migx ep fp8 int4 #78
Open
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
Map things to int8 right now as we don't explicitly set an int4 input type and pack/unpack int4 operands
TedThemistokleous
force-pushed
the
add_migx_ep_fp8_int4
branch
from
9b44fcc to
7076f74
Compare
TedThemistokleous
force-pushed
the
add_migx_ep_fp8_int4
branch
from
4850dfd to
11ff644
Compare
Mirror the same calibration code we use for int8 and just change which quantize we call through the MIGraphx API
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Outdated suggestions cannot be applied.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.
Description
Datatype support for int4 and fp8 (all formats)
Motivation and Context
Allows us to support operators of these data types to be handled by Onnxruntime MIGraphX EP