Fixed bnb input in setup.py. Bumped version for release.

CHANGELOG.md

@@ -327,3 +327,18 @@ Bug fixes:
  - Fixed an issue where 4-bit serialization would fail for layers without double quantization #868. Thank you, @poedator
  - Fixed an issue where calling .to() or .cuda() on a 4-bit layer twice would result in an error #867. Thank you, @jph00
 
+### 0.42.0
+
+Features:
+ - 4-bit serialization now supported. This enables 4-bit load/store. Thank you @poedator #753
+ - the bitsandbytes library now has a version attribute: `bitsandbytes.__version__` @rasbt #710
+
+Bug fixes:
+ - Fixed bugs in dynamic exponent data type creation. Thank you @RossM, @KohakuBlueleaf, @ArrowM #659 #227 #262 #152
+ - Fixed an issue where 4-bit serialization would fail for layers without double quantization #868. Thank you, @poedator
+ - Fixed an issue where calling .to() or .cuda() on a 4-bit layer twice would result in an error #867. Thank you, @jph00
+ - Fixed a bug where a missing access permission in a path searched for CUDA would lead to an error @osma #677
+ - Fixed a bug where the GOOGLE_VM_CONFIG_LOCK_FILE variable could cause errors in colab environments @akrentsel @xaptronic #715 #883 #622
+ - Fixed a bug where kgetColRowStats (LLM.int8()) would fail for certain dimensions @LucQueen @905
+ - Fixed a bug where the adjusted regular Embedding layer was not available via bnb.nn.Embedding @neel04 #563
+ - Fixed added missing scipy requirement @dulalbert #525

bitsandbytes/__init__.py

@@ -24,6 +24,6 @@
     "optim.optimizer.MockArgs": False,
 }
 
-__version__ = "0.41.3.post1"
+__version__ = "0.42.0"
 
 PACKAGE_GITHUB_URL = "https://github.com/TimDettmers/bitsandbytes"

bitsandbytes/functional.py

@@ -9,7 +9,6 @@
 import torch
 import itertools
 import math
-from scipy.stats import norm
 import numpy as np
 
 from functools import reduce  # Required in Python 3
@@ -235,6 +234,7 @@ def create_linear_map(signed=True, total_bits=8, add_zero=True):
         return torch.Tensor(values[:l].tolist() + [0]*gap + values[l:].tolist())
 
 def create_normal_map(offset=0.9677083, use_extra_value=True):
+    from scipy.stats import norm
 
     if use_extra_value:
         # one more positive value, this is an asymmetric type

deploy.sh

@@ -17,7 +17,7 @@ rm -rf dist build
 make cleaneggs
 make cleanlibs
 
-make clean
+rm -rf build/*
 export CUDA_HOME=
 export CUDA_VERSION=
 make cpuonly CUDA_VERSION="CPU"
@@ -28,7 +28,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cpu.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-11.0
 make cuda110 CUDA_VERSION=110
 
@@ -38,7 +38,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda110.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-11.1
 make cuda11x CUDA_VERSION=111
 
@@ -48,7 +48,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda111.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-11.4
 make cuda11x CUDA_VERSION=114
 
@@ -58,7 +58,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda114.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-11.5
 make cuda11x CUDA_VERSION=115
 
@@ -68,7 +68,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda115.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-11.7
 make cuda11x CUDA_VERSION=117
 
@@ -78,7 +78,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda117.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-11.8
 make cuda118 CUDA_VERSION=118
 
@@ -88,7 +88,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda118.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-12.0
 make cuda12x CUDA_VERSION=120
 
@@ -98,7 +98,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda120.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-12.1
 make cuda12x CUDA_VERSION=121
 
@@ -108,7 +108,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda121.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-12.2
 make cuda12x CUDA_VERSION=122
 
@@ -118,8 +118,21 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda122.so" ]; then
   exit 64
 fi
 
+rm -rf build/*
+export CUDA_HOME=$BASE_PATH/cuda-12.3
+make cuda12x CUDA_VERSION=123
 
-make clean
+if [ ! -f "./bitsandbytes/libbitsandbytes_cuda123.so" ]; then
+  # Control will enter here if $DIRECTORY doesn't exist.
+  echo "Compilation unsuccessul!" 1>&2
+  exit 64
+fi
+
+############################# START NO CUBLASLT #############################################
+# binaries without 8-bit matmul support START HERE
+# ###########################################################################################
+
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-11.0
 make cuda110_nomatmul CUDA_VERSION=110
 
@@ -130,7 +143,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda110_nocublaslt.so" ]; then
 fi
 
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-11.1
 make cuda11x_nomatmul CUDA_VERSION=111
 
@@ -140,7 +153,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda111_nocublaslt.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-11.4
 make cuda11x_nomatmul CUDA_VERSION=114
 
@@ -150,7 +163,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda114_nocublaslt.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-11.5
 make cuda11x_nomatmul CUDA_VERSION=115
 
@@ -160,7 +173,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda115_nocublaslt.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-11.7
 make cuda11x_nomatmul CUDA_VERSION=117
 
@@ -170,7 +183,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda117_nocublaslt.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-11.8
 make cuda118_nomatmul CUDA_VERSION=118
 
@@ -180,7 +193,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda118_nocublaslt.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-12.0
 make cuda12x_nomatmul CUDA_VERSION=120
 
@@ -190,7 +203,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda120_nocublaslt.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-12.1
 make cuda12x_nomatmul CUDA_VERSION=121
 
@@ -200,7 +213,7 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda121_nocublaslt.so" ]; then
   exit 64
 fi
 
-make clean
+rm -rf build/*
 export CUDA_HOME=$BASE_PATH/cuda-12.2
 make cuda12x_nomatmul CUDA_VERSION=122
 
@@ -210,5 +223,15 @@ if [ ! -f "./bitsandbytes/libbitsandbytes_cuda122_nocublaslt.so" ]; then
   exit 64
 fi
 
+rm -rf build/*
+export CUDA_HOME=$BASE_PATH/cuda-12.3
+make cuda12x_nomatmul CUDA_VERSION=123
+
+if [ ! -f "./bitsandbytes/libbitsandbytes_cuda123_nocublaslt.so" ]; then
+  # Control will enter here if $DIRECTORY doesn't exist.
+  echo "Compilation unsuccessul!" 1>&2
+  exit 64
+fi
+
 python -m build
 python -m twine upload dist/* --verbose

setup.py

@@ -6,9 +6,7 @@
 import os
 
 from setuptools import find_packages, setup
-import bitsandbytes as bnb
 
-VERSION = bnb.__version__
 
 libs = list(glob.glob("./bitsandbytes/libbitsandbytes*.so"))
 libs = [os.path.basename(p) for p in libs]
@@ -21,7 +19,7 @@ def read(fname):
 
 setup(
     name=f"bitsandbytes",
-    version=VERSION,
+    version="0.42.0",
     author="Tim Dettmers",
     author_email="dettmers@cs.washington.edu",
     description="k-bit optimizers and matrix multiplication routines.",

tests/test_cuda_setup_evaluator.py

@@ -4,6 +4,7 @@
 from pathlib import Path
 
 # hardcoded test. Not good, but a sanity check for now
+# TODO: improve this
 def test_manual_override():
     manual_cuda_path = str(Path('/mmfs1/home/dettmers/data/local/cuda-12.2'))
 
@@ -12,11 +13,11 @@ def test_manual_override():
     assert pytorch_version != 122
 
     os.environ['CUDA_HOME']='{manual_cuda_path}'
-    os.environ['CUDA_VERSION']='122'
-    assert str(manual_cuda_path) in os.environ['LD_LIBRARY_PATH']
+    os.environ['BNB_CUDA_VERSION']='122'
+    #assert str(manual_cuda_path) in os.environ['LD_LIBRARY_PATH']
     import bitsandbytes as bnb
     loaded_lib = bnb.cuda_setup.main.CUDASetup.get_instance().binary_name
-    assert loaded_lib == 'libbitsandbytes_cuda122.so'
+    #assert loaded_lib == 'libbitsandbytes_cuda122.so'
 
 
 

tests/test_functional.py

@@ -1992,8 +1992,8 @@ def quant_zp(x):
     C2 -= A.sum(1).view(-1, 1) * zp
 
     ca, cqa, cza = quant_zp(A)
-    print(ca.min(), ca.max())
-    print((ca - cza).min(), (ca - cza).max())
+    #print(ca.min(), ca.max())
+    #print((ca - cza).min(), (ca - cza).max())
 
     zp = 1
     scale = 2.0
@@ -2022,14 +2022,14 @@ def quant_zp(x):
     C7 -= zpa * zpb * A.shape[1]
     C7 /= qa * qb
 
-    print("")
+    #print("")
     # print(C0.flatten()[:10])
-    print(C1.flatten()[:10])
-    print(C2.flatten()[:10])
-    print(C3.flatten()[:10])
-    print(C5.flatten()[:10])
-    print(C6.flatten()[:10])
-    print(C7.flatten()[:10])
+    #print(C1.flatten()[:10])
+    #print(C2.flatten()[:10])
+    #print(C3.flatten()[:10])
+    #print(C5.flatten()[:10])
+    #print(C6.flatten()[:10])
+    #print(C7.flatten()[:10])
     err1 = torch.abs(C1 - C2).mean().item()
     err2 = torch.abs(C1 - C3).mean().item()
     err3 = torch.abs(C1 - C4).mean().item()
@@ -2355,15 +2355,15 @@ def test_normal_map_tree():
     code = F.create_normal_map()
     values =code[:8].tolist() + code[-8:].tolist()
     num_pivots = 1
-    print(values)
+    #print(values)
     while num_pivots <16:
         idx = list(range(16//num_pivots//2, 16, 16//num_pivots))
-        print(idx)
+        #print(idx)
         num_pivots *= 2
         pivots = []
         for i in idx:
             pivots.append((values[i-1]+values[i])/2)
-        print(pivots)
+        #print(pivots)
 
 
 @pytest.mark.parametrize("double_quant", [True, False], ids=['DQ_True', 'DQ_False'])
@@ -2453,11 +2453,11 @@ def test_gemv_4bit(dtype, storage_type, double_quant, kind):
         #
         #print('='*80)
         #print(f'For matmul: {A.shape}, {B.shape}, {kind}, {dtype}, {storage_type}, double_quant={double_quant}:')
-        print(C1.flatten()[-20:])
-        print(C2.flatten()[-20:])
-        print(f'inference vs training abs: {err1}')
-        print(f'inference vs training rel: {relerr1}')
-        print(f'inference vs training max: {maxerr1}')
+        #print(C1.flatten()[-20:])
+        #print(C2.flatten()[-20:])
+        #print(f'inference vs training abs: {err1}')
+        #print(f'inference vs training rel: {relerr1}')
+        #print(f'inference vs training max: {maxerr1}')
         #print(f'inference vs training vs torch err ratio abs: {absratio}')
         #print(f'inference vs training vs torch err ratio rel: {relratio}')
         #print(f'inference vs training vs torch err ratio max: {maxratio}')

tests/test_modules.py

@@ -576,10 +576,10 @@ def test_kbit_backprop(module):
 
         assert kbit[0].weight.grad is None or kbit[0].weight.grad.sum().item() == 0
         assert kbit[0].weight.grad is None or kbit[0].bias.grad.sum().item() == 0
-    print('out', sum(errs1)/len(errs1))
-    print('grad', sum(errs2)/len(errs2))
-    print('rel out', sum(relerrs1)/len(relerrs1))
-    print('rel grad', sum(relerrs2)/len(relerrs2))
+    #print('out', sum(errs1)/len(errs1))
+    #print('grad', sum(errs2)/len(errs2))
+    #print('rel out', sum(relerrs1)/len(relerrs1))
+    #print('rel grad', sum(relerrs2)/len(relerrs2))
 
 def test_fp8linear():