[dependabot]

Bumps the python-dependencies group with 9 updates in the / directory:

Package	From	To
pillow	12.3.0	13.0.0
click	8.4.2	8.5.0
python-dotenv	1.2.3	1.2.4
anthropic	1.0.0	1.15.0
openai	3.3.1	3.35.1
google-genai	2.19.0	2.31.0
torch	2.13.0	2.14.1
transformers	5.15.1	5.20.1
accelerate	1.14.0	1.15.0

Updates pillow from 12.3.0 to 13.0.0

Release notes

Sourced from pillow's releases.

13.0.0

https://pillow.readthedocs.io/en/stable/releasenotes/13.0.0.html

Removals

• Remove undocumented GifImagePlugin._save_netpbm #9901 [@​akx]
• Remove ExifTags.IFD.Makernote #9850 [@​radarhere]
• Drop support for Python 3.10 #9750 [@​hugovk]
• Remove deprecations for Pillow 13.0.0 #9751 [@​radarhere]

Deprecations

• Deprecate IM image format #9849 [@​radarhere]
• Deprecate ImageQt align8to32() in favour of raw encoder stride #9940 [@​akx]
• Deprecate JpegImageFile.load_djpeg #9924 [@​akx]

Documentation

• Update release notes for Pillow 13.0.0 #10223 [@​hugovk]
• Correct the ImageOps.fit centering corner documentation #10118 [@​FanWu-ai]
• Update Tidelift links #10102 [@​radarhere]
• Correct bullet point syntax in release notes #10069 [@​radarhere]
• Update ImageDraw signatures in docs #10065 [@​ZainnQureshii]
• Update macOS tested Pillow versions #10048 [@​hugovk]
• Add pillow-wmf and pillow-zx-spectrum to third-party plugins list #10037 [@​bitplane]
• Delete CLAUDE.md #10027 [@​hugovk]
• Add pillow-lunasvg to third-party plugins list #10016 [@​Maksim-Burtsev]
• Extend security policy #10018 [@​hugovk]
• Update Color3DLUT description #10004 [@​radarhere]
• Document the internal image distinctness constraints #9752 [@​akx]
• Update macOS tested Pillow versions #10000 [@​radarhere]
• Link security policy in project metadata #9980 [@​hugovk]
• Image.new(): correct docstring and comment about uninitialized memory #9960 [@​akx]
• Add AGENTS.md #9582 [@​aclark4life]
• Add release notes for #8811 #9937 [@​radarhere]
• Fixed typo #9923 [@​radarhere]
• Fixed typo #9922 [@​radarhere]
• C interface is internal #9879 [@​radarhere]
• Correct function references #9895 [@​radarhere]
• Revert move from python-pillow.org to python-pillow.github.io #9830 [@​aclark4life]
• Add additional CVE numbers to 12.3.0 release notes #9777 [@​radarhere]
• Remove remaining :rtype:s from docstrings #9742 [@​akx]
• Update macOS tested Pillow versions #9759 [@​radarhere]

Dependencies

• Update dependency cibuildwheel to v4.3.1 #10175 [@renovate[bot]]
• Update dependency cibuildwheel to v4.3.0 #10132 [@renovate[bot]]
• Update dependency harfbuzz to v14.5.0 #10108 [@renovate[bot]]
• Update github-actions #10111 [@renovate[bot]]

... (truncated)

Commits

• 293218a 13.0.0 version bump
• 262e8e4 Update release notes for Pillow 13.0.0 (#10223)
• af56ace Validate TIFF tag values against libtiff's own field definition before TIFFSe...
• 570df03 Validate TIFF tag values against libtiff's own field definition before TIFFSe...
• 53318b5 Take libtiff scanline geometry tags from the image, not tiffinfo/exif (#10226)
• 0a87417 Bound retained TIFF tag data by stream size (#10225)
• 11582e5 Fix heap buffer overflow in _tiffWriteProc when saving TIFF to BytesIO (#10224)
• 8501adc Fix test
• ec5f1cb Take libtiff scanline geometry tags from the image, not tiffinfo/exif
• 4b63dff Share TIFF tag data budget across frames
• Additional commits viewable in compare view

Updates click from 8.4.2 to 8.5.0

Release notes

Sourced from click's releases.

8.5.0

This is the Click 8.5.0 feature release. A feature release may include new features, remove previously deprecated code, add new deprecation, or introduce potentially breaking changes.

We encourage everyone to upgrade. You can read more about our Version Support Policy on our website.

PyPI: https://pypi.org/project/click/8.5.0/ Changes: https://click.palletsprojects.com/page/changes/#version-8-5-0 Milestone https://github.com/pallets/click/milestone/33

• Add built-in shell completion support for PowerShell (Windows PowerShell 5.1+ and pwsh 7+) alongside the existing bash, zsh, and fish completers. Use _FOO_BAR_COMPLETE=powershell_source foo-bar to generate the completion script. #2672 #3637
• Supported versions of Windows enable ANSI terminal styles by default. Colorama is no longer a dependency and is not used. #2986 #3505
• {class}Argument accepts a help parameter, and help output includes a Positional arguments section when argument help is available. #2983 #3473
• confirm() and prompt() strip ANSI color and style codes from the prompt when the output stream does not support them, matching echo(). This stripping was lost in 8.4.0 when #2969 began writing the prompt with input() directly. #3572 #3653
• {class}Path with allow_dash=True no longer triggers a BytesWarning, an error under python -bb, when checking a value against the - convention. #2877 #3642
• Add {func}custom_version_option, a --version option whose output is produced by a callback, covering cases {func}version_option intentionally does not. The feature set of {func}version_option is now frozen; see [discussion #3527](https://github.com/pallets/click/discussions/3527). #3581
• style() and secho() no longer silently drop the 256-color index 0 (black) passed as fg or bg, and now validate color arguments. Invalid colors raise a ValueError instead of a TypeError. #3677
• The automatic help option stores its value under the reserved name _click_default_help instead of help, so a parameter named help no longer breaks parsing. The new name is visible in {meth}Command.to_info_dict output. Parameters that overwrite each other's value trigger a warning: an argument sharing its name with another parameter, or any parameter claiming the reserved name. Options may still share a name to compete for the same value (feature switches). #2819 #3678
• unstyle and the ANSI handling behind help-text wrapping now strip the full CSI escape-sequence grammar. #3681
• Streamline Option flag handling: the flag-kind, type, lazy-default and validation steps in Option.__init__ move into focused helpers, and flag_value and default keep their unset sentinel at construction (resolved lazily on read) so is UNSET reliably tells a user-supplied value from an auto-derived one. Runtime behavior is unchanged, but {meth}Parameter.to_info_dict now resolves default=True on a feature switch to its flag_value, matching what the function receives at call

... (truncated)

Changelog

Sourced from click's changelog.

Version 8.5.0

Released 2026-08-24

• Add built-in shell completion support for PowerShell (Windows PowerShell 5.1+ and pwsh 7+) alongside the existing bash, zsh, and fish completers. Use _FOO_BAR_COMPLETE=powershell_source foo-bar to generate the completion script. {issue}2672 {pr}3637
• Supported versions of Windows enable ANSI terminal styles by default. Colorama is no longer a dependency and is not used. {issue}2986 {pr}3505
• {class}Argument accepts a help parameter, and help output includes a Positional arguments section when argument help is available. {issue}2983 {pr}3473
• confirm() and prompt() strip ANSI color and style codes from the prompt when the output stream does not support them, matching echo(). This stripping was lost in 8.4.0 when {pr}2969 began writing the prompt with input() directly. {issue}3572 {pr}3653
• Fix test failures when using pytest >= 9.1. {pr}3656
• {class}Path with allow_dash=True no longer triggers a BytesWarning, an error under python -bb, when checking a value against the - convention. {issue}2877 {pr}3642
• Add {func}custom_version_option, a --version option whose output is produced by a callback, covering cases {func}version_option intentionally does not. The feature set of {func}version_option is now frozen; see [discussion #3527](https://github.com/pallets/click/discussions/3527). {pr}3581
• style() and secho() no longer silently drop the 256-color index 0 (black) passed as fg or bg, and now validate color arguments. Invalid colors raise a ValueError instead of a TypeError. {pr}3677
• The automatic help option stores its value under the reserved name _click_default_help instead of help, so a parameter named help no longer breaks parsing. The new name is visible in {meth}Command.to_info_dict output. Parameters that overwrite each other's value trigger a warning: an argument sharing its name with another parameter, or any parameter claiming the reserved name. Options may still share a name to compete for the same value (feature switches). {issue}2819 {pr}3678
• unstyle and the ANSI handling behind help-text wrapping now strip the full CSI escape-sequence grammar. {pr}3681
• Streamline Option flag handling: the flag-kind, type, lazy-default and validation steps in Option.__init__ move into focused helpers, and flag_value and default keep their unset sentinel at construction (resolved lazily on read) so is UNSET reliably tells a user-supplied value from an auto-derived one. Runtime behavior is unchanged, but {meth}Parameter.to_info_dict now resolves default=True on a feature switch to its flag_value, matching what the function receives at call time. {pr}3641
• {func}get_binary_stream and {func}get_text_stream are deprecated and will be removed in Click 9.0. {issue}3481 {pr}3695
• The following click.utils names were never intentionally public and are now private (_-prefixed). The old names remain available with a DeprecationWarning until Click 9.0: LazyFile, KeepOpenFile,

... (truncated)

Commits

• 8b19813 Release version 8.5.0
• 2c8cd3a Add FAQ entry about UnicodeEncodeError on Windows (#3778)
• 131c86a Add FAQ entry about UnicodeEncodeError on Windows
• e1fd594 Add support of pathlib.Path to edit (#3781)
• a1d8785 Add support of pathlib.Path to edit
• 2103e15 Forward all user's parameters set in PAGER and improve flag detection (#3777)
• a6256bf Forwards all user's parameters set in PAGER
• 61b69e9 Resolve the pager command once, in _pager_contextmanager (#3776)
• 9835b0f Resolve the pager command once, in _pager_contextmanager
• f36d58b Refactor pager stream handling (#3767)
• Additional commits viewable in compare view

Updates python-dotenv from 1.2.3 to 1.2.4

Release notes

Sourced from python-dotenv's releases.

v1.2.4

Fixed

• dotenv get no longer exits with code 1 for empty string values (KEY=) by [@​ShamikOfficial] in #700
• An unquoted empty value followed by an inline comment (e.g. KEY= # comment) is now parsed as an empty string instead of the comment text by [@​Noethix55555] in #663
• dotenv run --no-override now expands variable references with the same precedence as load_dotenv(override=False), so a value like ${BASE}/suffix uses the existing BASE from the environment instead of the one from the .env file by [@​ROTl24] in #698

Changelog

Sourced from python-dotenv's changelog.

[1.2.4] - 2026-10-01

Fixed

• dotenv get no longer exits with code 1 for empty string values (KEY=) by [@​ShamikOfficial] in #700
• An unquoted empty value followed by an inline comment (e.g. KEY= # comment) is now parsed as an empty string instead of the comment text by [@​Noethix55555] in #663
• dotenv run --no-override now expands variable references with the same precedence as load_dotenv(override=False), so a value like ${BASE}/suffix uses the existing BASE from the environment instead of the one from the .env file by [@​ROTl24] in #698

Commits

• a565c2c Bump version: 1.2.3 → 1.2.4
• 4a7abd0 docs: add 1.2.4 release notes (#663, #698, #700)
• f215c02 fix: dotenv get exits 0 for empty string values (#700)
• 58f2d7c test: make test_run_with_command_flags portable and meaningful (#709)
• e0310e5 fix: honor --no-override when expanding variables in dotenv run (#698)
• a00cb2e docs: add CHANGELOG entry for #663 (fix #600)
• f5485a6 fix: parse empty unquoted value with inline comment as empty string
• See full diff in compare view

Updates anthropic from 1.0.0 to 1.15.0

Release notes

Sourced from anthropic's releases.

v1.15.0

Features

• api: add between_tools to thinking types in model capabilities
• api: add the speed parameter and usage.speed to Messages
• api: return a Console organization's usage-tier spend cap as a read-only spend limit

Chores

• ci: speed up breaking change detection
• docs: describe headers on Managed Agents and telemetry endpoints
• internal: simplify the script that formats code blocks in docs
• tests: cover OpenTelemetry options and method default timeouts

Documentation

• api: note that some models need a beta header to use speed

v1.14.0

Features

• api: add execution_identity to Managed Agents agents

• api: add organization_type_not_supported error to spend limit lists

• api: add telemetry destination test endpoint

• api: add telemetry destinations to the Admin API

• client: add native support for OpenTelemetry tracing

  If your app registers a global tracer provider, the SDK will create spans with it after upgrading, and they'll be exported with the rest of your traces. Requests will also get trace context headers from the global propagator. To disable, pass open_telemetry=False to the client or set ANTHROPIC_OPEN_TELEMETRY=false. Learn more in the documentation.

Bug Fixes

• api: rename telemetry destination test response model
• client: close dropped streams without waiting for the collector
• streaming: pass unknown event types through the message stream helpers
• streaming: stop raising "Unexpected event order" for an event before message_start

Chores

• docs: correct prefill, thinking, fallback and model descriptions
• docs: document the scope required to assign the managed role
• docs: update model examples
• internal: narrow the tripwire test rule in the contributor guides

Documentation

... (truncated)

Changelog

Sourced from anthropic's changelog.

1.15.0 (2026-10-15)

Features

• api: add between_tools to thinking types in model capabilities
• api: add the speed parameter and usage.speed to Messages
• api: return a Console organization's usage-tier spend cap as a read-only spend limit

Chores

• ci: speed up breaking change detection
• docs: describe headers on Managed Agents and telemetry endpoints
• internal: simplify the script that formats code blocks in docs
• tests: cover OpenTelemetry options and method default timeouts

Documentation

• api: note that some models need a beta header to use speed

1.14.0 (2026-10-14)

Features

• api: add execution_identity to Managed Agents agents

• api: add organization_type_not_supported error to spend limit lists

• api: add telemetry destination test endpoint

• api: add telemetry destinations to the Admin API

• client: add native support for OpenTelemetry tracing

  If your app registers a global tracer provider, the SDK will create spans with it after upgrading, and they'll be exported with the rest of your traces. Requests will also get trace context headers from the global propagator. To disable, pass open_telemetry=False to the client or set ANTHROPIC_OPEN_TELEMETRY=false. Learn more in the documentation.

Bug Fixes

• api: rename telemetry destination test response model
• client: close dropped streams without waiting for the collector
• streaming: pass unknown event types through the message stream helpers
• streaming: stop raising "Unexpected event order" for an event before message_start

Chores

• docs: correct prefill, thinking, fallback and model descriptions
• docs: document the scope required to assign the managed role
• docs: update model examples
• internal: narrow the tripwire test rule in the contributor guides

... (truncated)

Commits

• bcd55fe release: 1.15.0
• 6974344 Merge pull request #1021 from anthropics/synced/published
• 80404f6 chore(tests): cover OpenTelemetry options and method default timeouts
• ef91336 feat(helpers): retry a rate-limited fast fallback at standard speed (#1024)
• 9ace6c5 feat(api): add between_tools to thinking types in model capabilities
• d44d218 chore(docs): describe headers on Managed Agents and telemetry endpoints
• ef59b87 docs(api): note that some models need a beta header to use speed
• c10fe77 docs(changelog): shorten the OpenTelemetry tracing note (#1023)
• 5b18741 chore(internal): simplify the script that formats code blocks in docs
• f0e30dc feat(api): add the speed parameter and usage.speed to Messages
• Additional commits viewable in compare view

Updates openai from 3.3.1 to 3.35.1

Release notes

Sourced from openai's releases.

v3.35.1

3.35.1 (2026-10-18)

Bug Fixes

• preserve beta Realtime byte reads on websockets 13 (#4102) (5766349)

v3.35.0

3.35.0 (2026-10-17)

Features

• api: add typed Live sideband events (#4099) (8d5cbf0)
• api: add ultrafast service tier for Agents (#4101) (100bf8f)

Bug Fixes

• realtime: preserve byte receives with websockets 13 (#4098) (309916e)

v3.34.0

3.34.0 (2026-10-17)

Features

• api: add degraded external storage status (#4097) (b216cb2)
• api: add Realtime Translation WebSocket sessions (#4095) (bfada50)

v3.33.0

3.33.0 (2026-10-16)

Features

• api: add safety identifiers to image requests and responses (#4093) (c969e32)
• api: share Decisions cache breakpoint parameter type (#4092) (f7cefda)

v3.32.0

3.32.0 (2026-10-16)

Features

• api: add cache settings and key options [2/2] (#4087) (325da45)
• api: add prompt-based voice creation (#4091) (2e39db7)
• api: add regional processing residency options (#4081) (21ce0c2)
• api: add response error diagnostics and model notices (#4086) (e71d781)

... (truncated)

Changelog

Sourced from openai's changelog.

3.35.1 (2026-10-18)

Bug Fixes

• preserve beta Realtime byte reads on websockets 13 (#4102) (5766349)

3.35.0 (2026-10-17)

Features

• api: add typed Live sideband events (#4099) (8d5cbf0)
• api: add ultrafast service tier for Agents (#4101) (100bf8f)

Bug Fixes

• realtime: preserve byte receives with websockets 13 (#4098) (309916e)

3.34.0 (2026-10-17)

Features

• api: add degraded external storage status (#4097) (b216cb2)
• api: add Realtime Translation WebSocket sessions (#4095) (bfada50)

3.33.0 (2026-10-16)

Features

• api: add safety identifiers to image requests and responses (#4093) (c969e32)
• api: share Decisions cache breakpoint parameter type (#4092) (f7cefda)

3.32.0 (2026-10-16)

Features

• api: add cache settings and key options [2/2] (#4087) (325da45)
• api: add prompt-based voice creation (#4091) (2e39db7)
• api: add regional processing residency options (#4081) (21ce0c2)
• api: add response error diagnostics and model notices (#4086) (e71d781)
• api: support audio fixture playback in Live sessions (#4085) (9998c1a)
• api: support Decisions in Live sessions (#4088) (230a923)

Bug Fixes

... (truncated)

Commits

• 547ecb1 release: 3.35.1 (#4103)
• 5766349 fix: preserve beta Realtime byte reads on websockets 13 (#4102)
• ab91400 release: 3.35.0 (#4100)
• 100bf8f feat(api): add ultrafast service tier for Agents (#4101)
• 8d5cbf0 feat(api): add typed Live sideband events (#4099)
• 309916e fix(realtime): preserve byte receives with websockets 13 (#4098)
• 3eb0837 release: 3.34.0 (#4096)
• b216cb2 feat(api): add degraded external storage status (#4097)
• bfada50 feat(api): add Realtime Translation WebSocket sessions (#4095)
• b7c5e5d release: 3.33.0 (#4094)
• Additional commits viewable in compare view

Updates google-genai from 2.19.0 to 2.31.0

Release notes

Sourced from google-genai's releases.

v2.31.0

2.31.0 (2026-10-14)

Features

• interactions: Add gemini-4-argon to Model enum (8554b07)

Bug Fixes

• Clear continuation state and preserve prompt feedback when continuation hop returns no candidates (a0d809c)
• Fix the function declaration parser for union types. Fixes #3060 (be32e9e)
• Fix the grounding metadata merging for automatic continuation in generate content and generate content stream (f1499bb)
• Make continuation streams retry dropped hops, stop on prompt_feedback, and keep real errors in generate_content (339a929)
• Match the mTLS endpoint to each transport's SSL context, including plain aiohttp (f8a4693)

v2.30.1

2.30.1 (2026-10-13)

Bug Fixes

• Concatenate output_text across multi-step model responses in a turn (180c345)

v2.30.0

2.30.0 (2026-10-13)

Features

• Accumulate streaming continuation usage metadata and expose initial_prompt_token_count (b27f3cd)
• Add continuation_token to the Interactions interaction.status_update event. (02db4c5)
• Add a CONTINUATION_REQUIRED Interaction status (bf9d6d7)
• Add billable token counts fields (6a6501c)
• Add startOffset/endOffset to Transcription and interimTranscriptTimestampEnabled to RealtimeInputConfig (c644a94)
• Add usage in InteractionStatusUpdate events (56c3a1a)
• Allow automatic resumption of continuation token for in generate content and generate content stream. Default behavior for this feature is disabled (83e43ea)
• cli: Update Speakeasy generator to v2.948.0, bump CLI to v0.4.2, and apply CLI overlay fixes (cdcd3dd)
• Enable automatic continuation token resumption for chats module. (9ebf714)

Bug Fixes

• Clear intermediate CONTINUATION finish_reason in streaming continuation (8a641d8)
• Do not record incomplete continuation stream turns in curated chat history (b382e01)
• gemini-api-cli: Remove unsupported 2.5 and 1.6 models from Interactions model options and CLI examples (fe5fb10)
• genai: Deduplicate Flash and Omni descriptions in ModelOption (33efaad)
• interactions: Restore super().model_construct in Interaction model validator (ccd5dd3)
• python: Discard stale checkpoint token on non-resumable finish reason (f44e8ee)

... (truncated)

Changelog

Sourced from google-genai's changelog.

2.31.0 (2026-10-14)

Features

• interactions: Add gemini-4-argon to Model enum (8554b07)

Bug Fixes

• Clear continuation state and preserve prompt feedback when continuation hop returns no candidates (a0d809c)
• Fix the function declaration parser for union types. Fixes #3060 (be32e9e)
• Fix the grounding metadata merging for automatic continuation in generate content and generate content stream (f1499bb)
• Make continuation streams retry dropped hops, stop on prompt_feedback, and keep real errors in generate_content (339a929)
• Match the mTLS endpoint to each transport's SSL context, including plain aiohttp (f8a4693)

2.30.1 (2026-10-13)

Bug Fixes

• Concatenate output_text across multi-step model responses in a turn (180c345)

2.30.0 (2026-10-13)

Features

• Accumulate streaming continuation usage metadata and expose initial_prompt_token_count (b27f3cd)
• Add continuation_token to the Interactions interaction.status_update event. (02db4c5)
• Add a CONTINUATION_REQUIRED Interaction status (bf9d6d7)
• Add billable token counts fields (6a6501c)
• Add startOffset/endOffset to Transcription and interimTranscriptTimestampEnabled to RealtimeInputConfig (c644a94)
• Add usage in InteractionStatusUpdate events (56c3a1a)
• Allow automatic resumption of continuation token for in generate content and generate content stream. Default behavior for this feature is disabled (83e43ea)
• cli: Update Speakeasy generator to v2.948.0, bump CLI to v0.4.2, and apply CLI overlay fixes (cdcd3dd)
• Enable automatic continuation token resumption for chats module. (9ebf714)

Bug Fixes

• Clear intermediate CONTINUATION finish_reason in streaming continuation (8a641d8)
• Do not record incomplete continuation stream turns in curated chat history (b382e01)
• gemini-api-cli: Remove unsupported 2.5 and 1.6 models from Interactions model options and CLI examples (fe5fb10)
• genai: Deduplicate Flash and Omni descriptions in ModelOption (33efaad)
• interactions: Restore super().model_construct in Interaction model validator (ccd5dd3)
• python: Discard stale checkpoint token on non-resumable finish reason (f44e8ee)
• Raise default HTTP timeouts so long-running requests aren't killed client-side (b006b4b)
• Resume streaming generation from the latest checkpoint continuation token on mid-stream errors or cutoffs (90bb32b)
• Update EnvironmentConfig.env schema to map keys to EnvVar or string. (3108d8c)

... (truncated)

Commits

• 6bbead9 chore(main): release 2.31.0 (#3123)
• be32e9e fix: fix the function declaration parser for union types. Fixes #3060
• efbe9d7 chore: internal change
• 339a929 fix: make continuation streams retry dropped hops, stop on prompt_feedback, a...
• 829ec55 test: retry transient errors for longer in the nightly integration tests
• 1dbf070 test: switch the shared integration tests to gemini-3.8-flash
• f1499bb fix: fix the grounding metadata merging for automatic continuation in generat...
• 8554b07 feat(interactions): add gemini-4-argon to Model enum
• a0d809c fix: clear continuation state and preserve prompt feedback when continuation ...
• f8a4693 fix: Match the mTLS endpoint to each transport's SSL context, including plain...
• Additional commits viewable in compare view

Updates torch from 2.13.0 to 2.14.1

Release notes

Sourced from torch's releases.

PyTorch 2.14.1 Release

This release is meant to fix the following regressions and silent correctness issues:

Silent correctness fixes

• Fix incorrect torch.linalg.lstsq solutions on MPS for complex batched underdetermined systems (#196113), fixed by #196128
• Fix non-orthogonal U and inaccurate small singular values from torch.linalg.svd on MPS for rank-deficient and ill-conditioned inputs (#196112), fixed by #196139 and #199063
• Update the CUDA 13.2 Linux binaries to CUDA 13.2.2 (#196351). This NVIDIA update resolves two critical issues that could produce incorrect results (CUDA 13.2.2 release notes):
  • cuBLAS: cublasLtMatmul() could ignore tensor-wide scaling for NVFP4 matrix multiplications (introduced in CUDA 13.2 Update 1)
  • Compiler: failed thread reconvergence could leave stale or corrupted register values in kernels with nested thread divergence (present since CUDA 12.8)

Regression fixes

• Fix torch.linalg.svd, torch.linalg.svdvals and torch.linalg.lstsq failing on MPS with a Metal pipeline-state error for inputs above 8192 elements (#195937), fixed by #195949 and #195950
• Fix internal assert in torch.svd(out=) on MPS for complex inputs (#195822), fixed by #195872

PyTorch 2.14.0 Release Notes

• Highlights
• Backwards Incompatible Changes
• Deprecations
• New Features
• Improvements
• Bug fixes
• Performance
• Documentation
• Security
• Developers

Highlights

For more details about these highlighted features, you can look at the release blogpost. Below are the full release notes for this release.

<...

Description has been truncated