GitLab

* Change 8bit optimizer blocksize 2048->256; additional bf16 support
* Update tolerances for 8bit optimizer tests

* Add AdEMAMix optimizer

* Add PagedAdEMAMix32bit, AdEMAMix32bit

* Add PagedAdEMAMix32bit, AdEMAMix32bit

* AdEMAMix: add support for alpha/beta3 scheduling

* Update paged AdEMAMix

This makes it possible to load an optimizer checkpoint without
automatically moving the optimizer's state to the GPU.

Updates the requirements on [matplotlib](https://github.com/matplotlib/matplotlib) to permit the latest version.

Updates `matplotlib` to 3.9.2
- [Release notes](https://github.com/matplotlib/matplotlib/releases)
- [Commits](https://github.com/matplotlib/matplotlib/compare/v3.9.1...v3.9.2

)

---
updated-dependencies:
- dependency-name: matplotlib
  dependency-type: direct:development
  dependency-group: major
...
Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Updates the requirements on [pytest](https://github.com/pytest-dev/pytest), [wheel](https://github.com/pypa/wheel) and [scipy](https://github.com/scipy/scipy) to permit the latest version.

Updates `pytest` from 8.3.2 to 8.3.3
- [Release notes](https://github.com/pytest-dev/pytest/releases)
- [Changelog](https://github.com/pytest-dev/pytest/blob/main/CHANGELOG.rst)
- [Commits](https://github.com/pytest-dev/pytest/compare/8.3.2...8.3.3)

Updates `wheel` to 0.44.0
- [Release notes](https://github.com/pypa/wheel/releases)
- [Changelog](https://github.com/pypa/wheel/blob/main/docs/news.rst)
- [Commits](https://github.com/pypa/wheel/compare/0.43.0...0.44.0)

Updates `scipy` from 1.14.0 to 1.14.1
- [Release notes](https://github.com/scipy/scipy/releases)
- [Commits](https://github.com/scipy/scipy/compare/v1.14.0...v1.14.1

)

---
updated-dependencies:
- dependency-name: pytest
  dependency-type: direct:production
  update-type: version-update:semver-patch
  dependency-group: minor-patch
- dependency-name: wheel
  dependency-type: direct:development
  dependency-group: minor-patch
- dependency-name: scipy
  dependency-type: direct:production
  update-type: version-update:semver-patch
  dependency-group: minor-patch
...
Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Update for VS2022 17.11 compatibility with CUDA < 12.4

* Try again

* remove kcompress

* fix initial template call

* fix function name

* remove vector load

* cleanup reduce  & rearrange

* format

Signed-off-by: mgiessing <marvin.giessing@gmail.com>

* Done

* fix format

* fix format

* fix format

* fix format

* Address format error and fix default arg bug

* Refine stream argument passing mechanism

* Fix bug

* Delete unused code

* Embedding4bit and Embedding8bit implementation

* lint

* Update bitsandbytes/nn/modules.py
Co-authored-by: Matthew Douglas <38992547+matthewdouglas@users.noreply.github.com>

* Update bitsandbytes/nn/modules.py
Co-authored-by: Matthew Douglas <38992547+matthewdouglas@users.noreply.github.com>

* Update bitsandbytes/nn/modules.py
Co-authored-by: Matthew Douglas <38992547+matthewdouglas@users.noreply.github.com>

* saving -> Saving

---------
Co-authored-by: Matthew Douglas <38992547+matthewdouglas@users.noreply.github.com>

[FSDP] Enable loading prequantized weights with bf16/fp16/fp32 quant_storage

Bumps the minor-patch group with 1 update: [pytest](https://github.com/pytest-dev/pytest).


Updates `pytest` from 8.3.1 to 8.3.2
- [Release notes](https://github.com/pytest-dev/pytest/releases)
- [Changelog](https://github.com/pytest-dev/pytest/blob/main/CHANGELOG.rst)
- [Commits](https://github.com/pytest-dev/pytest/compare/8.3.1...8.3.2

)

---
updated-dependencies:
- dependency-name: pytest
  dependency-type: direct:production
  update-type: version-update:semver-patch
  dependency-group: minor-patch
...
Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Fix restoration of quant_storage for CPU offloading

* Clarify comment on default quant_storage in Params4bit.from_prequantized()

* fix to make quant_storage dynamic based on serialized dtype

* delete obsolete comment

---------
Co-authored-by: Titus von Koeller <9048635+Titus-von-Koeller@users.noreply.github.com>

Bumps the minor-patch group with 1 update: [pytest](https://github.com/pytest-dev/pytest).


Updates `pytest` from 8.2.2 to 8.3.1
- [Release notes](https://github.com/pytest-dev/pytest/releases)
- [Changelog](https://github.com/pytest-dev/pytest/blob/main/CHANGELOG.rst)
- [Commits](https://github.com/pytest-dev/pytest/compare/8.2.2...8.3.1

)

---
updated-dependencies:
- dependency-name: pytest
  dependency-type: direct:production
  update-type: version-update:semver-minor
  dependency-group: minor-patch
...
Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Fixed optim update error with non-contiguous grads
* fix formatting

Thanks @Edenzzzz for this contribution!

---------
Co-authored-by: Titus von Koeller <9048635+Titus-von-Koeller@users.noreply.github.com>

* fix min_8bit_size invalid bug

* Apply same fix to other optimizer base class

---------
Co-authored-by: Edenzzzz <wtan45@wisc.edu>

* Add CUDA 12.5 builds and enable CUDA 12.4 on Windows

* Update install doc

* chore: delete useless buffered activation

* fix: fix bugs

* fixed test_4bit_warnings on cpu-only platforms

* fixed linear8bit-based tests for cpu only platforms

Updates the requirements on [matplotlib](https://github.com/matplotlib/matplotlib) to permit the latest version.

Updates `matplotlib` to 3.9.1
- [Release notes](https://github.com/matplotlib/matplotlib/releases)
- [Commits](https://github.com/matplotlib/matplotlib/compare/v3.9.0...v3.9.1

)

---
updated-dependencies:
- dependency-name: matplotlib
  dependency-type: direct:development
  dependency-group: major
...
Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Add support for building with latest MSVC

* Update MSVC 1940+ support for CUDA builds.