Format merged branch

Author: Qubitium

gptqmodel/nn_modules/qlinear/marlin_awq.py

@@ -20,16 +20,16 @@
     apply_awq_marlin_linear,
     awq_marlin_repack,
     awq_to_marlin_zero_points,
-    marlin_supports_fp8_input_capability,
-    marlin_int4_fp8_preprocess,
     marlin_import_exception,
+    marlin_int4_fp8_preprocess,
     marlin_make_empty_g_idx,
     marlin_make_workspace_new,
     marlin_permute_bias,
     marlin_permute_scales,
     marlin_runtime_available,
     marlin_runtime_error,
     marlin_supports_fp8_input,
+    marlin_supports_fp8_input_capability,
     replace_parameter,
 )
 from ...utils.marlin_scalar_type import scalar_types

gptqmodel/quantization/input_activations_triton.py

@@ -7,6 +7,7 @@
 
 from ..utils.env import env_flag
 
+
 try:
     import triton
     import triton.language as tl

tests/kernels/test_selection.py

@@ -14,9 +14,9 @@
 from gptqmodel.nn_modules.qlinear.machete import MacheteLinear
 from gptqmodel.nn_modules.qlinear.machete_awq import AwqMacheteLinear
 from gptqmodel.nn_modules.qlinear.marlin_awq import AwqMarlinLinear
-from gptqmodel.nn_modules.qlinear.torch_awq import AwqTorchLinear
 from gptqmodel.nn_modules.qlinear.torch_aten_kernel import TorchAtenLinear
 from gptqmodel.nn_modules.qlinear.torch_aten_kernel_awq import TorchAtenAwqLinear
+from gptqmodel.nn_modules.qlinear.torch_awq import AwqTorchLinear
 from gptqmodel.quantization import FORMAT, METHOD
 from gptqmodel.utils import importer
 from gptqmodel.utils.backend import BACKEND

tests/models/test_llama3_2_awq_w4a8_fp8_dynamic.py

@@ -8,7 +8,6 @@
 
 import pytest
 import torch
-
 from model_test import ModelTest
 
 from gptqmodel import BACKEND

tests/test_awq_input_activations.py

@@ -8,14 +8,14 @@
 from torch import nn
 
 import gptqmodel.looper.awq_processor as awq_processor_module
-from gptqmodel.nn_modules.qlinear.torch_awq import AwqTorchLinear
 from gptqmodel.models.writer import QUANT_ACT_LOG_TYPE
+from gptqmodel.nn_modules.qlinear.torch_awq import AwqTorchLinear
 from gptqmodel.quantization import FORMAT, METHOD
 from gptqmodel.quantization.config import QuantizeConfig
 from gptqmodel.quantization.input_activations import (
     calibrate_input_scale_inv,
-    quantize_input,
     quantize_dequantize_input,
+    quantize_input,
 )
 from gptqmodel.quantization.input_activations_triton import supports_triton_fp8_input_quant
 

tests/test_split_by_layer_save.py

@@ -9,7 +9,6 @@
 from accelerate import load_checkpoint_in_model
 
 from gptqmodel.models.writer import (
-    ModelWriter,
     PROCESS_LOG_LAYER,
     PROCESS_LOG_MODULE,
     PROCESS_LOG_NAME,
@@ -20,6 +19,7 @@
     QUANT_ACT_LOG_TYPE,
     QUANT_LOG_LOSS,
     QUANT_LOG_NSAMPLES,
+    ModelWriter,
 )
 from gptqmodel.quantization.config import FORMAT, METHOD
 

tests/test_stage_modules.py

@@ -6,11 +6,11 @@
 import torch
 
 import gptqmodel.looper.stage_subset as stage_subset_module
+from gptqmodel.looper.forward_executor import ForwardExecutor
 from gptqmodel.looper.loop_processor import ExecutionConfig
 from gptqmodel.looper.module_looper import FinalizeProgressInfo, ModuleLooper
 from gptqmodel.looper.named_module import NamedModule
 from gptqmodel.looper.paroquant_processor import ParoQuantProcessor
-from gptqmodel.looper.forward_executor import ForwardExecutor
 from gptqmodel.looper.stage_inputs_capture import StageInputsCapture
 from gptqmodel.looper.stage_layer import (
     _capture_pristine_group_context,
@@ -307,7 +307,7 @@ def _run_executor_single(executor, processor, *, apply_moe_config):
 def _run_executor_parallel(executor, processor, *, apply_moe_config):
     def clone_module_for_devices_fn(module, devices, progress_callback=None):
         del progress_callback
-        return {device: module for device in devices}
+        return dict.fromkeys(devices, module)
 
     def forward_batch_worker_fn(
         _replica,