[CORE] Consolidate 6+ kernel boolean toggels args to single Backend arg #68
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
Qubitium
reviewed
Jun 26, 2024
Qubitium
changed the title
DeJoker
pushed a commit
to DeJoker/GPTQModel
that referenced
this pull request
Jul 19, 2024
DeJoker
pushed a commit
to DeJoker/GPTQModel
that referenced
this pull request
Jul 19, 2024
* Consolidate Backend * change Backend.TRITON_V2 to Backend.TRITON * According to quantize_config.format, determine when the Backend is packing the model. * Auto choose the fastest one Backend based on quant model compatibility * fix issue: Automatically select Backend, returns incorrect qlinear. * cleanup * cleanup
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Outdated suggestions cannot be applied.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.
Resolves #59
The following args will be merged into single backed: Backend = Backend.AUTO
use_triton: bool,
disable_exllama: bool = False,
disable_exllamav2: bool = False,
use_marlin: bool = False,
use_bitblas: bool = True,
Reason: It is not only super confusing for users to use correctly (matrix condition of passive binary toggles), even project developers ran into multiple bugs due to these passive switches. We can't keep adding more binary toggles every time we add a backend/kernel/runtime. Becoming unmaintainable and unusable by both end-users and project devs.
Prelim design:
class Backend(ENUM):
AUTO # choose the fastest one based on quant model compatibility
CUDA_OLD
CUDA
TRITON_V2
EXLLAMA
EXLLAMA_V2
MARLIN
BITBLAS