Add support for compressed-tensors w8a8 int checkpoints #2745
Conversation
This file contains hidden or 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
This change adds a loader for w8a8 int checkpoints. One large benefit of int8 support is that the corresponding cutlass matmul kernels also work on compute capability 7.5. Evaluation on neuralmagic/Meta-Llama-3.1-8B-Instruct-quantized.w8a8: | Tasks |Version| Filter |n-shot| Metric | |Value | |Stderr| |---------------|------:|----------------|-----:|-----------------------|---|-----:|---|------| |gsm8k_cot_llama| 3|flexible-extract| 8|exact_match |↑ |0.8431|± |0.0100| | | |strict-match | 8|exact_match |↑ |0.8393|± |0.0101| |ifeval | 4|none | 0|inst_level_loose_acc |↑ |0.8597|± | N/A| | | |none | 0|inst_level_strict_acc |↑ |0.8201|± | N/A| | | |none | 0|prompt_level_loose_acc |↑ |0.7967|± |0.0173| | | |none | 0|prompt_level_strict_acc|↑ |0.7468|± |0.0187| Which is the same ballpark as vLLM. As usual, lots of thanks to Neural Magic/vLLM for the kernels.
danieldk
force-pushed
the
feature/compressed-tensors-w8a8-int
branch
from
bdb6d0a to
b2dc10a
Compare
It's far less flaky and gives better output.
drbh
previously approved these changes
Nov 18, 2024
server/text_generation_server/layers/compressed_tensors/w8a8_int.py
Outdated
Show resolved
Hide resolved
Co-authored-by: drbh <[email protected]>
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Outdated suggestions cannot be applied.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.
What does this PR do?
This change adds a loader for w8a8 int checkpoints. One large benefit of int8 support is that the corresponding cutlass matmul kernels also work on compute capability 7.5.
Evaluation on neuralmagic/Meta-Llama-3.1-8B-Instruct-quantized.w8a8:
Which is the same ballpark as vLLM.
Note: output of static input quantization with zero points (asymmetric) seems really bad. But it's hard to tell if it's a bug in this PR or just static input quantization just being bad in general (for the expected reasons). vLLM also doesn't give good output in this case, though it seems slightly better? (E.g. on
nm-testing/asym-w8w8-int8-static-per-tensor-tiny-llama.)As usual, lots of thanks to Neural Magic/vLLM for the kernels.
Before submitting
Pull Request section?
to it if that's the case.
documentation guidelines, and
here are tips on formatting docstrings.
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.