[ Misc ] Refactor Marlin Python Utilities (#6082)

Co-authored-by: Robert Shaw <rshaw@neuralmagic.com>
This commit is contained in:
Robert Shaw
2024-07-11 15:40:11 +00:00
committed by GitHub
co-authored by Robert Shaw
parent 55f692b46e
commit b675069d74
12 changed files with 690 additions and 728 deletions
+6 -4
View File
@@ -5,14 +5,16 @@ import torch.utils.benchmark as benchmark
from benchmark_shapes import WEIGHT_SHAPES
from vllm import _custom_ops as ops
from vllm.model_executor.layers.quantization.gptq_marlin import (
GPTQ_MARLIN_MAX_PARALLEL, GPTQ_MARLIN_MIN_THREAD_N,
GPTQ_MARLIN_SUPPORTED_GROUP_SIZES, GPTQ_MARLIN_SUPPORTED_NUM_BITS)
from vllm.model_executor.layers.quantization.gptq_marlin_24 import (
GPTQ_MARLIN_24_MAX_PARALLEL, GPTQ_MARLIN_24_MIN_THREAD_N,
GPTQ_MARLIN_24_SUPPORTED_GROUP_SIZES, GPTQ_MARLIN_24_SUPPORTED_NUM_BITS)
from vllm.model_executor.layers.quantization.utils.marlin_utils import (
MarlinWorkspace, marlin_24_quantize, marlin_quantize)
GPTQ_MARLIN_MAX_PARALLEL, GPTQ_MARLIN_MIN_THREAD_N,
GPTQ_MARLIN_SUPPORTED_GROUP_SIZES, GPTQ_MARLIN_SUPPORTED_NUM_BITS)
from vllm.model_executor.layers.quantization.utils.marlin_utils_test import (
MarlinWorkspace, marlin_quantize)
from vllm.model_executor.layers.quantization.utils.marlin_utils_test_24 import (
marlin_24_quantize)
from vllm.model_executor.layers.quantization.utils.quant_utils import (
gptq_pack, quantize_weights, sort_weights)
from vllm.utils import FlexibleArgumentParser