(WIP) Multi platform abstraction tweaks #1195
Closed
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
|
I was thinking about this a little bit in #1173 actually. It may be good opportunity to rename some of these ops to make them more clear, even if they still are aliased for BC reasons. Possible examples: |
| raise NotImplementedError | ||
|
|
||
|
|
||
| class FourBitMatmul(ABC): |
There was a problem hiding this comment.
Something to think about here is that for the nested quantization, the interface for KBitQuantization is needed (quantize_blockwise/dequantize_blockwise)
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
Merged
|
Closing this as it will be replaced by upcoming torch.library refactor |
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Outdated suggestions cannot be applied.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.
I'll fill in more details as I go along here, but for now I'm just publicly sharing what I'm working on. Feel free to already comment if anything catches your eye.
I've gathered extensive feedback from previous PRs (some closed) around this topic and am implementing my interpretation of what's needed, what Tim wants (API structure/ sub-interfaces) and what the community suggested, e.g. tensor-driven dispatch + deferred initialization.
The idea is that I'll ask for extensive feedback from the community once I'm through with the changes I'm imagining (aiming for this week) and we see how to continue from there and then merge the Intel PR #1178, once it's been adapted to this slightly modified architecture.