Ruff fixes (#984)

* Adjust Ruff configuration

* do not autofix always
* be less strict around tests and benchmarks
* adjust ignores for now

* Ruff: autofix I and F401

* Apply ruff autofixes

* Fix RUF013 complaint

* Fix mutable default in replace_linear

* Don't use bare except

* Wrap bitsandbytes.__main__ entrypoint in function; fix "sensible" typo

* Fix ruff B008 (function call in arguments)

* Add ruff noqas as suitable

* Fix RUF005 (splat instead of concatenating)

* Fix B018 (useless expression)

* Add pre-commit configuration + GitHub Actions lint workflow

* Fix unused `e` in bitsandbytes/__main__.py

* fix merge conflict resolution error

* run pre-commit hook

---------

Co-authored-by: Titus <[email protected]>

Author: akx

.github/workflows/lint.yml

@@ -0,0 +1,19 @@
+name: Lint
+
+on:
+  push:
+    branches:
+      - main
+  pull_request:
+
+jobs:
+  Lint:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v4
+        with:
+          python-version: "3.12"
+      - uses: pre-commit/[email protected]
+        env:
+          RUFF_OUTPUT_FORMAT: github

.pre-commit-config.yaml

@@ -0,0 +1,8 @@
+repos:
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.1.15
+    hooks:
+      - id: ruff
+        args:
+          - --fix
+      # - id: ruff-format  # TODO: enable when the time is right

benchmarking/switchback/make_plot_with_jsonl.py

@@ -1,9 +1,7 @@
-import matplotlib.pyplot as plt
-import pandas as pd
-import numpy as np
-import os
 
 import matplotlib.gridspec as gridspec
+import matplotlib.pyplot as plt
+import pandas as pd
 
 cmap=plt.get_cmap('cool')
 

benchmarking/switchback/speed_benchmark.py

@@ -1,14 +1,22 @@
 import json
-
 import time
+
 import torch
-import torch.nn as nn
 
+from bitsandbytes.triton.int8_matmul_mixed_dequantize import (
+    int8_matmul_mixed_dequantize,
+)
+from bitsandbytes.triton.int8_matmul_rowwise_dequantize import (
+    int8_matmul_rowwise_dequantize,
+)
+from bitsandbytes.triton.quantize_columnwise_and_transpose import (
+    quantize_columnwise_and_transpose,
+)
+from bitsandbytes.triton.quantize_global import (
+    quantize_global,
+    quantize_global_transpose,
+)
 from bitsandbytes.triton.quantize_rowwise import quantize_rowwise
-from bitsandbytes.triton.quantize_columnwise_and_transpose import quantize_columnwise_and_transpose
-from bitsandbytes.triton.int8_matmul_rowwise_dequantize import int8_matmul_rowwise_dequantize
-from bitsandbytes.triton.quantize_global import quantize_global, quantize_global_transpose
-from bitsandbytes.triton.int8_matmul_mixed_dequantize import int8_matmul_mixed_dequantize
 
 # KNOW ISSUE: need to optimize "w_quantize_colwise_transpose" when embeddim is too large.
 

bitsandbytes/__init__.py

@@ -3,14 +3,14 @@
 # This source code is licensed under the MIT license found in the
 # LICENSE file in the root directory of this source tree.
 
-from . import cuda_setup, utils, research
+from . import cuda_setup, research, utils
 from .autograd._functions import (
     MatmulLtState,
     bmm_cublas,
     matmul,
+    matmul_4bit,
     matmul_cublas,
     mm_cublas,
-    matmul_4bit
 )
 from .cextension import COMPILED_WITH_CUDA
 from .nn import modules

bitsandbytes/__main__.py

@@ -1,11 +1,7 @@
 import os
+from os.path import isdir
 import sys
-import shlex
-import subprocess
-
 from warnings import warn
-from typing import Tuple
-from os.path import isdir
 
 import torch
 
@@ -20,7 +16,7 @@ def find_file_recursive(folder, filename):
             out = glob.glob(os.path.join(folder, "**", filename + ext))
             outs.extend(out)
     except Exception as e:
-        raise RuntimeError('Error: Something when wrong when trying to find file. {e}')
+        raise RuntimeError('Error: Something when wrong when trying to find file.') from e
 
     return outs
 
@@ -62,14 +58,11 @@ def generate_bug_report_information():
                     print_header(f"{path} CUDA PATHS")
                     paths = find_file_recursive(path, '*cuda*')
                     print(paths)
-            except:
-                print(f'Could not read LD_LIBRARY_PATH: {path}')
+            except Exception as e:
+                print(f'Could not read LD_LIBRARY_PATH: {path} ({e})')
     print('')
 
 
-
-
-
 def print_header(
     txt: str, width: int = HEADER_WIDTH, filler: str = "+"
 ) -> None:
@@ -78,67 +71,61 @@ def print_header(
 
 
 def print_debug_info() -> None:
+    from . import PACKAGE_GITHUB_URL
     print(
         "\nAbove we output some debug information. Please provide this info when "
         f"creating an issue via {PACKAGE_GITHUB_URL}/issues/new/choose ...\n"
     )
 
 
-generate_bug_report_information()
+def main():
+    generate_bug_report_information()
 
+    from . import COMPILED_WITH_CUDA
+    from .cuda_setup.main import get_compute_capabilities
 
-from . import COMPILED_WITH_CUDA, PACKAGE_GITHUB_URL
-from .cuda_setup.env_vars import to_be_ignored
-from .cuda_setup.main import get_compute_capabilities
-
+    print_header("OTHER")
+    print(f"COMPILED_WITH_CUDA = {COMPILED_WITH_CUDA}")
+    print(f"COMPUTE_CAPABILITIES_PER_GPU = {get_compute_capabilities()}")
+    print_header("")
+    print_header("DEBUG INFO END")
+    print_header("")
+    print("Checking that the library is importable and CUDA is callable...")
+    print("\nWARNING: Please be sure to sanitize sensitive info from any such env vars!\n")
 
-print_header("OTHER")
-print(f"COMPILED_WITH_CUDA = {COMPILED_WITH_CUDA}")
-print(f"COMPUTE_CAPABILITIES_PER_GPU = {get_compute_capabilities()}")
-print_header("")
-print_header("DEBUG INFO END")
-print_header("")
-print(
-    """
-Running a quick check that:
-    + library is importable
-    + CUDA function is callable
-"""
-)
-print("\nWARNING: Please be sure to sanitize sensible info from any such env vars!\n")
+    try:
+        from bitsandbytes.optim import Adam
 
-try:
-    from bitsandbytes.optim import Adam
+        p = torch.nn.Parameter(torch.rand(10, 10).cuda())
+        a = torch.rand(10, 10).cuda()
 
-    p = torch.nn.Parameter(torch.rand(10, 10).cuda())
-    a = torch.rand(10, 10).cuda()
+        p1 = p.data.sum().item()
 
-    p1 = p.data.sum().item()
+        adam = Adam([p])
 
-    adam = Adam([p])
+        out = a * p
+        loss = out.sum()
+        loss.backward()
+        adam.step()
 
-    out = a * p
-    loss = out.sum()
-    loss.backward()
-    adam.step()
+        p2 = p.data.sum().item()
 
-    p2 = p.data.sum().item()
+        assert p1 != p2
+        print("SUCCESS!")
+        print("Installation was successful!")
+    except ImportError:
+        print()
+        warn(
+            f"WARNING: {__package__} is currently running as CPU-only!\n"
+            "Therefore, 8-bit optimizers and GPU quantization are unavailable.\n\n"
+            f"If you think that this is so erroneously,\nplease report an issue!"
+        )
+        print_debug_info()
+    except Exception as e:
+        print(e)
+        print_debug_info()
+        sys.exit(1)
 
-    assert p1 != p2
-    print("SUCCESS!")
-    print("Installation was successful!")
-    sys.exit(0)
 
-except ImportError:
-    print()
-    warn(
-        f"WARNING: {__package__} is currently running as CPU-only!\n"
-        "Therefore, 8-bit optimizers and GPU quantization are unavailable.\n\n"
-        f"If you think that this is so erroneously,\nplease report an issue!"
-    )
-    print_debug_info()
-    sys.exit(0)
-except Exception as e:
-    print(e)
-    print_debug_info()
-    sys.exit(1)
+if __name__ == "__main__":
+    main()

bitsandbytes/autograd/__init__.py

@@ -1 +1 @@
-from ._functions import undo_layout, get_inverse_transform_indices
+from ._functions import get_inverse_transform_indices, undo_layout

bitsandbytes/autograd/_functions.py

@@ -1,8 +1,8 @@
-import operator
-import warnings
 from dataclasses import dataclass
 from functools import reduce  # Required in Python 3
-from typing import Tuple, Optional, Callable
+import operator
+from typing import Callable, Optional, Tuple
+import warnings
 from warnings import warn
 
 import torch

bitsandbytes/cextension.py

@@ -1,12 +1,9 @@
 import ctypes as ct
-import os
-import torch
-
-from pathlib import Path
 from warnings import warn
 
-from bitsandbytes.cuda_setup.main import CUDASetup
+import torch
 
+from bitsandbytes.cuda_setup.main import CUDASetup
 
 setup = CUDASetup.get_instance()
 if setup.initialized != True:
@@ -25,7 +22,7 @@
         Inspect the output of the command and see if you can locate CUDA libraries. You might need to add them
         to your LD_LIBRARY_PATH. If you suspect a bug, please take the information from python -m bitsandbytes
         and open an issue at: https://github.com/TimDettmers/bitsandbytes/issues''')
-    lib.cadam32bit_grad_fp32 # runs on an error if the library could not be found -> COMPILED_WITH_CUDA=False
+    _ = lib.cadam32bit_grad_fp32  # runs on an error if the library could not be found -> COMPILED_WITH_CUDA=False
     lib.get_context.restype = ct.c_void_p
     lib.get_cusparse.restype = ct.c_void_p
     lib.cget_managed_ptr.restype = ct.c_void_p

bitsandbytes/cuda_setup/main.py

@@ -17,15 +17,15 @@
 """
 
 import ctypes as ct
-import os
 import errno
+import os
+from pathlib import Path
 import platform
-import torch
+from typing import Set, Union
 from warnings import warn
-from itertools import product
 
-from pathlib import Path
-from typing import Set, Union
+import torch
+
 from .env_vars import get_potentially_lib_path_containing_env_vars
 
 # these are the most common libs names
@@ -111,14 +111,16 @@ def manual_override(self):
         if torch.cuda.is_available():
             if 'BNB_CUDA_VERSION' in os.environ:
                 if len(os.environ['BNB_CUDA_VERSION']) > 0:
-                    warn((f'\n\n{"="*80}\n'
-                          'WARNING: Manual override via BNB_CUDA_VERSION env variable detected!\n'
-                          'BNB_CUDA_VERSION=XXX can be used to load a bitsandbytes version that is different from the PyTorch CUDA version.\n'
-                          'If this was unintended set the BNB_CUDA_VERSION variable to an empty string: export BNB_CUDA_VERSION=\n'
-                          'If you use the manual override make sure the right libcudart.so is in your LD_LIBRARY_PATH\n'
-                          'For example by adding the following to your .bashrc: export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:<path_to_cuda_dir/lib64\n'
-                          f'Loading CUDA version: BNB_CUDA_VERSION={os.environ["BNB_CUDA_VERSION"]}'
-                          f'\n{"="*80}\n\n'))
+                    warn(
+                        f'\n\n{"=" * 80}\n'
+                        'WARNING: Manual override via BNB_CUDA_VERSION env variable detected!\n'
+                        'BNB_CUDA_VERSION=XXX can be used to load a bitsandbytes version that is different from the PyTorch CUDA version.\n'
+                        'If this was unintended set the BNB_CUDA_VERSION variable to an empty string: export BNB_CUDA_VERSION=\n'
+                        'If you use the manual override make sure the right libcudart.so is in your LD_LIBRARY_PATH\n'
+                        'For example by adding the following to your .bashrc: export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:<path_to_cuda_dir/lib64\n'
+                        f'Loading CUDA version: BNB_CUDA_VERSION={os.environ["BNB_CUDA_VERSION"]}'
+                        f'\n{"=" * 80}\n\n'
+                    )
                     binary_name = self.binary_name.rsplit(".", 1)[0]
                     suffix = ".so" if os.name != "nt" else ".dll"
                     self.binary_name = binary_name[:-3] + f'{os.environ["BNB_CUDA_VERSION"]}.{suffix}'
@@ -207,7 +209,7 @@ def remove_non_existent_dirs(candidate_paths: Set[Path]) -> Set[Path]:
         try:
             if path.exists():
                 existent_directories.add(path)
-        except PermissionError as pex:
+        except PermissionError:
             # Handle the PermissionError first as it is a subtype of OSError 
             # https://docs.python.org/3/library/exceptions.html#exception-hierarchy
             pass
@@ -217,8 +219,10 @@ def remove_non_existent_dirs(candidate_paths: Set[Path]) -> Set[Path]:
 
     non_existent_directories: Set[Path] = candidate_paths - existent_directories
     if non_existent_directories:
-        CUDASetup.get_instance().add_log_entry("The following directories listed in your path were found to "
-            f"be non-existent: {non_existent_directories}", is_warning=False)
+        CUDASetup.get_instance().add_log_entry(
+            f"The following directories listed in your path were found to be non-existent: {non_existent_directories}",
+            is_warning=False,
+        )
 
     return existent_directories
 
@@ -360,8 +364,10 @@ def evaluate_cuda_setup():
     cuda_version_string = get_cuda_version()
 
     cuda_setup.add_log_entry(f"CUDA SETUP: PyTorch settings found: CUDA_VERSION={cuda_version_string}, Highest Compute Capability: {cc}.")
-    cuda_setup.add_log_entry(f"CUDA SETUP: To manually override the PyTorch CUDA version please see:"
-                             "https://github.com/TimDettmers/bitsandbytes/blob/main/how_to_use_nonpytorch_cuda.md")
+    cuda_setup.add_log_entry(
+        "CUDA SETUP: To manually override the PyTorch CUDA version please see:"
+        "https://github.com/TimDettmers/bitsandbytes/blob/main/how_to_use_nonpytorch_cuda.md"
+    )
 
 
     # 7.5 is the minimum CC vor cublaslt