address review comments

Author: prigoyal

docs/source/framework/pytorch_integration/debugging.rst

@@ -16,7 +16,7 @@ can use these flags to enable logging. Various types of flags exposed are:
 * :code:`debug_tuner`: print debug spew for the tuner multithreading behavior.
 
 
-In order to use enable these flags, you need to call :code:`tc.GlobalDebugInit`
+In order to use enable these flags, you need to call :code:`tc.SetDebugFlags`
 and set the proper flags to :code:`true`. All of these flags are :code:`boolean`
 flags that take values :code:`true` or :code:`false`.
 
@@ -28,14 +28,14 @@ Example usage
     import tensor_comprehensions as tc
     import torch
 
-    tc.GlobalDebugInit(debug_tc_mapper=True, debug_lang=False)
+    tc.SetDebugFlags(debug_tc_mapper=True, debug_lang=False)
 
     matmul = tc.define(tc.database['matmul']['lang'], name='matmul')
     mat1, mat2 = torch.randn(3, 4).cuda(), torch.randn(4, 5).cuda()
     out = matmul(mat1, mat2)
 
 In above example, when the TC executes, we will see the TC mapper information.
-You can chose to set any number of flags but the :code:`tc.GlobalDebugInit` should
+You can chose to set any number of flags but the :code:`tc.SetDebugFlags` should
 only be called once.
 
 Printing TC generated CUDA code
@@ -50,7 +50,7 @@ and the generated CUDA code will be printed on command line.
     import tensor_comprehensions as tc
     import torch
 
-    tc.GlobalDebugInit(dump_cuda=True)
+    tc.SetDebugFlags(dump_cuda=True)
 
     matmul = tc.define(tc.database['matmul']['lang'], name='matmul')
     mat1, mat2 = torch.randn(3, 4).cuda(), torch.randn(4, 5).cuda()

tensor_comprehensions/__init__.py

@@ -18,7 +18,7 @@
 from tensor_comprehensions.tc_unit import TcUnit
 from tensor_comprehensions.tc_unit import TcAutotuner
 from tensor_comprehensions.tc_unit import TcCompilationUnit
-from tensor_comprehensions.tc_unit import GlobalDebugInit
+from tensor_comprehensions.tc_unit import SetDebugFlags
 from tensor_comprehensions.tc_unit import autotuner_settings
 from tensor_comprehensions.tc_unit import small_sizes_autotuner_settings
 from tensor_comprehensions.tc_unit import ATenCompilationUnit
@@ -27,6 +27,6 @@
 
 __all__ = [
     'define', 'TcUnit', 'TcAutotuner', 'TcCompilationUnit', 'autotuner_settings',
-    'small_sizes_autotuner_settings', 'GlobalDebugInit', 'ATenCompilationUnit',
+    'small_sizes_autotuner_settings', 'SetDebugFlags', 'ATenCompilationUnit',
     'Options', 'database', 'decode',
 ]

tensor_comprehensions/tc_unit.py

@@ -41,7 +41,7 @@
 ###############################################################################
 # Set global debugging flags
 ###############################################################################
-class GlobalDebugInit(object):
+class SetDebugFlags(object):
     def __init__(self, **kwargs):
         self.set_gflags(**kwargs)
 

test_python/layers/test_dump_cuda.py

@@ -21,7 +21,7 @@
 
 # enable this to dump cuda code generated whenever tc layer runs: simple run or
 # autotuner run
-tc.GlobalDebugInit(dump_cuda=True)
+tc.SetDebugFlags(dump_cuda=True)
 
 
 class TestDumpCuda(unittest.TestCase):

test_python/layers/test_layernorm.py

@@ -19,7 +19,7 @@
 import torch.cuda
 import unittest
 
-tc.GlobalDebugInit(debug_tuner=False, debug_tc_mapper=False)
+tc.SetDebugFlags(debug_tuner=False, debug_tc_mapper=False)
 
 
 class TestLayerNorm(unittest.TestCase):

test_python/test_debug_init.py

@@ -19,7 +19,7 @@
 import torch.cuda
 import tensor_comprehensions as tc
 
-tc.GlobalDebugInit(dump_cuda=True, debug_tc_mapper=True)
+tc.SetDebugFlags(dump_cuda=True, debug_tc_mapper=True)
 
 
 class TestDebugInit(unittest.TestCase):

test_python/test_tc_torch.py

@@ -24,7 +24,7 @@
 from tensor_comprehensions.mapping_options import Options
 from common import TestCase, run_tests
 
-tc.GlobalDebugInit(dump_cuda=False)
+tc.SetDebugFlags(dump_cuda=False)
 
 
 MATMUL_LANG = """