[SW-177468] Removed unused code + cleanup

Change-Id: I4d27c067e87c1a30eb1da9df16a16c46d092c638

Author: HolyFalafel

neural_compressor/torch/algorithms/fp8_quant/_core/common.py

@@ -55,7 +55,6 @@ def __init__(self, num_inputs, param_names, num_outputs, required_output):
 }
 descale_fcn = lambda x, scale: torch.mul(x, scale)
 scale_fcn = lambda x, scale: torch.div(x, scale)
-mat_scale_fcn = lambda x, scale_col, scale_row: torch.div(torch.div(x, scale_col), scale_row)
 cast_fcn = lambda x, dtype: x.to(dtype=dtype)
 cast_to_fp8_fcn = lambda x, dtype, scale_inv=None: torch.ops.hpu.cast_to_fp8_v2(x, scale_inv, False, False, dtype)[0]
 cast_from_fp8_fcn = lambda x, dtype, scale=None: torch.ops.hpu.cast_from_fp8(x, scale, dtype)
@@ -76,25 +75,6 @@ def rec_fn(x, fn):
         return fn(x)
 
 
-def np_to_pt(x):
-    return rec_fn(x, lambda x: torch.tensor(x) if isinstance(x, np.ndarray) else x)
-
-
-def pt_to_np(x):
-    return rec_fn(
-        x,
-        lambda x: (x.detach().cpu().float().numpy() if isinstance(x, torch.Tensor) else x),
-    )
-
-
-def np_to_list(x):
-    return rec_fn(x, lambda x: x.tolist() if isinstance(x, np.ndarray) else x)
-
-
-def list_to_np(x):
-    return rec_fn(x, lambda x: np.array(x) if isinstance(x, list) else x)
-
-
 def save_json(d, fname):
     with open(fname, "w") as f:
         json.dump(d, f, indent=4)

neural_compressor/torch/algorithms/fp8_quant/_core/measure.py

@@ -9,6 +9,7 @@
     ScaleMethod,
     MeasureExclude,
     get_hqt_config,
+    set_hqt_config,
 )
 from .common import *
 from ..utils.logger import logger
@@ -229,11 +230,6 @@ def load_measurements(model, fname):
     return d
 
 
-def get_default_config(mod_list):
-    config = {k: "default" for k in mod_list}
-    return config
-
-
 def save_json(d, fname):
     with open(fname, "w") as f:
         json.dump(d, f, indent=4)

neural_compressor/torch/algorithms/fp8_quant/_core/scale.py

@@ -1,7 +1,7 @@
 import torch
 import numpy as np
 
-from .._quant_common.quant_config import ScaleMethod
+from .._quant_common.quant_config import ScaleMethod, set_hqt_config
 from .scale_methods import *
 from .quant_dequant import *
 

neural_compressor/torch/algorithms/fp8_quant/_quant_common/helper_modules.py

@@ -1,7 +1,7 @@
 import torch.nn as nn
 import torch
 
-from .quant_config import QuantMode, get_hqt_config, set_hqt_config
+from .quant_config import QuantMode, get_hqt_config
 
 try:  # backwards compatibility for 1.16
     from habana_frameworks.torch.hpex.kernels import fp8_fused_sdpa

neural_compressor/torch/algorithms/fp8_quant/_quant_common/quant_config.py

@@ -13,7 +13,6 @@
 
 local_rank = int(os.getenv("LOCAL_RANK", "-1"))
 world_size = int(os.getenv("WORLD_SIZE", "-1"))
-global_rank = int(os.getenv("RANK", "-1"))
 
 
 class QuantMode(Enum):

test/3x/torch/algorithms/fp8_quant/unit_tests/test_deepspeed.py

@@ -1,4 +1,3 @@
-import os
 import typing
 
 import pytest