-
-
Notifications
You must be signed in to change notification settings - Fork 4.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Hardware][CPU] Support AWQ for CPU backend #7515
base: main
Are you sure you want to change the base?
Conversation
👋 Hi! Thank you for contributing to the vLLM project. Once the PR is approved and ready to go, please make sure to run full CI as it is required to merge (or just use auto-merge). To run full CI, you can do one of these:
🚀 |
Thanks! |
Hi @mgoin , would you please help to review this PR? Because I added a platform filter in |
qconfig=qconfig, | ||
bias=bias, | ||
group_size=self.quant_config.group_size, | ||
quant_method=1 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Is there a more informative value that could be used here?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This argument is defined as int for now. I added a dict mapping to make it more informative. (AWQ=1, GPTQ=2)
weight_dtype = ipex.quantization.WoqWeightDtype.INT4 | ||
lowp_mode = ipex.quantization.WoqLowpMode.INT8 | ||
act_quant_mode = ipex.quantization.WoqActQuantMode.PER_BATCH |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can you describe these settings? It seems like this may be quantizing activations as well?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes, the lowp_mode
means the compute type
, using int8 for FMA instructions with better performance. The activation is dynamic per-token quantized to int8 with the setting PER_BATCH
.
Hi @mgoin , do you have any further comments about this? I think this PR is ready to go :) BTW, seems the CPU CI on the main branch is broken due to a new MOE model test and the lack of the AZP epilogue. I disabled the compressed-tensor test and will fix it recently. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This looks to be in a good place now, thanks! I just have a nit on UX @bigPYJ1151
Please let me know if I can assist with the compressed-tensors test as we will want to enable that again.
|
||
bias = layer.bias if not layer.skip_bias_add else None | ||
|
||
import intel_extension_for_pytorch as ipex |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Could you put a try-catch around the ipex import to provide a better instructive error message to the user? It is also important to enforce versioning if needed.
Something like this deepspeed backend
vllm/vllm/model_executor/layers/quantization/deepspeedfp.py
Lines 144 to 153 in 151ef4e
try: | |
import deepspeed | |
if deepspeed.__version__ < "0.14.2": | |
raise ImportError("deepspeed version is wrong. Please " | |
"install deepspeed>=0.14.2.") | |
from deepspeed.ops.fp_quantizer import FP_Quantize | |
except ImportError as err: | |
raise ImportError("Please install deepspeed>=0.14.2 via " | |
"`pip install deepspeed>=0.14.2` to use " | |
"deepspeedfp quantizer.") from err |
This PR added a new
ipex_quant
configuration to support various weight-only quantization method with IPEX for the CPU backend.AWQ is supported for now, a basic test is included.
PR Checklist (Click to Expand)
Thank you for your contribution to vLLM! Before submitting the pull request, please ensure the PR meets the following criteria. This helps vLLM maintain the code quality and improve the efficiency of the review process.
PR Title and Classification
Only specific types of PRs will be reviewed. The PR title is prefixed appropriately to indicate the type of change. Please use one of the following:
[Bugfix]
for bug fixes.[CI/Build]
for build or continuous integration improvements.[Doc]
for documentation fixes and improvements.[Model]
for adding a new model or improving an existing model. Model name should appear in the title.[Frontend]
For changes on the vLLM frontend (e.g., OpenAI API server,LLM
class, etc.)[Kernel]
for changes affecting CUDA kernels or other compute kernels.[Core]
for changes in the core vLLM logic (e.g.,LLMEngine
,AsyncLLMEngine
,Scheduler
, etc.)[Hardware][Vendor]
for hardware-specific changes. Vendor name should appear in the prefix (e.g.,[Hardware][AMD]
).[Misc]
for PRs that do not fit the above categories. Please use this sparingly.Note: If the PR spans more than one category, please include all relevant prefixes.
Code Quality
The PR need to meet the following code quality standards:
format.sh
to format your code.docs/source/
if the PR modifies the user-facing behaviors of vLLM. It helps vLLM user understand and utilize the new features or changes.Notes for Large Changes
Please keep the changes as concise as possible. For major architectural changes (>500 LOC excluding kernel/data/config/test), we would expect a GitHub issue (RFC) discussing the technical design and justification. Otherwise, we will tag it with
rfc-required
and might not go through the PR.What to Expect for the Reviews
The goal of the vLLM team is to be a transparent reviewing machine. We would like to make the review process transparent and efficient and make sure no contributor feel confused or frustrated. However, the vLLM team is small, so we need to prioritize some PRs over others. Here is what you can expect from the review process:
action-required
label on the PR if there are changes required. The contributor should address the comments and ping the reviewer to re-review the PR.Thank You
Finally, thank you for taking the time to read these guidelines and for your interest in contributing to vLLM. Your contributions make vLLM a great tool for everyone!