Commits

Commits on May 8, 2023

update README
Show description for e7b0c90
fa0311
committed
e7b0c90
Copy full SHA for e7b0c90
Merge branch 'TimDettmers-8bit_better_tests'
Show description for a85aaf8
fa0311
committed
a85aaf8
Copy full SHA for a85aaf8
Merge branch '8bit_better_tests' of https://github.com/TimDettmers/bitsandbytes into TimDettmers-8bit_better_tests
Show description for d063f11
fa0311
committed
d063f11
Copy full SHA for d063f11

Commits on Apr 20, 2023

Handle more cases in Linear8bitLt tests
mryab
committed
9e00b62
Copy full SHA for 9e00b62
Make Linear8bitLt serialization more memory-efficient
mryab
committed
2cf062a
Copy full SHA for 2cf062a

Commits on Apr 12, 2023

Added last SwitchBack refactors. All tests green.
TimDettmers
committed
9e7cdc9
Copy full SHA for 9e7cdc9
Added triton utils.
TimDettmers
committed
008dfff
Copy full SHA for 008dfff
Added is_available_triton guard to Triton SwitchBackLinear.
TimDettmers
committed
5b612bc
Copy full SHA for 5b612bc
Added is_available_triton guard.
TimDettmers
committed
c3d87e4
Copy full SHA for c3d87e4
Merge branch 'main' into fp8_merge
TimDettmers
committed
7140c01
Copy full SHA for 7140c01
Added missing example folder.
TimDettmers
committed
32f8c89
Copy full SHA for 32f8c89
Refactored simulated fp8 modules into research.nn.
TimDettmers
committed
dd562c2
Copy full SHA for dd562c2
Added missing triton and fp8 files.
TimDettmers
committed
e67bfcc
Copy full SHA for e67bfcc
Refactored triton into its own folder. Refactored fp8 matmuls.
TimDettmers
committed
ec1ea63
Copy full SHA for ec1ea63
Added better error message for debugging on CUDA not detected failures.
TimDettmers
committed
7c65101
Copy full SHA for 7c65101

Commits on Apr 11, 2023

Fixing #300.
TimDettmers
committed
659a7df
Copy full SHA for 659a7df
Updates README and CHANGELOG.
TimDettmers
committed
eb1c331
Copy full SHA for eb1c331
Added more detailed cuda setup debug and debugging instructions.
TimDettmers
committed
89e3b82
Copy full SHA for 89e3b82
Fixed CUDA Conda PyTorch 2.0 issues.
TimDettmers
committed
4cd63de
Copy full SHA for 4cd63de
Added pre/post call to all lib calls. Fixes #120
TimDettmers
committed
2bb5c00
Copy full SHA for 2bb5c00
Updated change log.
TimDettmers
committed
29ab3a6
Copy full SHA for 29ab3a6
Fixed bug where beta2 was not passed into Lion 32-bit.
TimDettmers
committed
2eb3108
Copy full SHA for 2eb3108
Fixed noisy tests for 8-bit Lion.
TimDettmers
committed
792af5c
Copy full SHA for 792af5c
Added launch bounds to fix launch resource error for Lion.
TimDettmers
committed
0b2ebcd
Copy full SHA for 0b2ebcd
Merge pull request #159 from TimDettmers/serialize_8bit
Show description for ed6f3eb
TimDettmers
authored
ed6f3eb
Copy full SHA for ed6f3eb
Merge pull request #188 from lucidrains/main
Show description for b0ec20c
TimDettmers
authored
b0ec20c
Copy full SHA for b0ec20c
Merge pull request #190 from svgsponer/Fix#157
Show description for d3e0e39
TimDettmers
authored
d3e0e39
Copy full SHA for d3e0e39
Merge pull request #213 from tonylins/dev/fix_no_absmax
Show description for c787553
TimDettmers
authored
c787553
Copy full SHA for c787553
Merge pull request #260 from rapsealk/fix_libsbitsandbytes_cpu_so
Show description for 6b4c5af
TimDettmers
authored
6b4c5af
Copy full SHA for 6b4c5af
Merge pull request #292 from justheuristic/patch-2
Show description for 72efa32
TimDettmers
authored
72efa32
Copy full SHA for 72efa32

Commits on Apr 10, 2023

Support 1650, 1660
justheuristic
authored
5e456be
Copy full SHA for 5e456be

Commits on Apr 8, 2023

typo
Mitchell Wortsman
committed
d677a71
Copy full SHA for d677a71
mem efficient"
Mitchell Wortsman
committed
da524d9
Copy full SHA for da524d9

Commits on Apr 5, 2023

Fixed typo libsbitsandbytes_cpu.so
rapsealk
committed
8cceff7
Copy full SHA for 8cceff7

Commits on Apr 1, 2023

clarify in readme
Mitchell Wortsman
committed
eb6c53c
Copy full SHA for eb6c53c