Bump the nuget group with 3 updates

[dependabot]

Updated FastBertTokenizer from 1.0.28 to 1.2.6.

Release notes

Sourced from FastBertTokenizer's releases.

1.2.6

Highlights

• New net10.0 target (#​134). The package now ships net10.0, net8.0 and netstandard2.0. On .NET 9+ the span-based vocabulary lookup uses FrozenDictionary.GetAlternateLookup instead of the unsafe char* key trick, so that target is compiled without unsafe code.
• added_tokens support (#​100). LoadTokenizerJson now honours the added_tokens section of tokenizer.json; if several added tokens match at a position, the longest one wins, like in Hugging Face (#​101, thanks @​kudima03).
• strip_accents support. tokenizer.json files that set normalizer.strip_accents explicitly (including false) now load. If it is unset, it follows lowercase, as in the original BERT.
• cleanup_tokenization_spaces from the configuration (#​105, #​107, thanks @​kudima03). Decode now uses the tokenizer's decoder.cleanup setting.
• Faster pre-tokenization (#​151). ASCII characters are classified with a lookup table, which cuts single-threaded encode time by roughly 6–9%.
• Bounded worst case for very long words (#​156). WordPiece matching now starts at the length of the longest vocabulary entry, not the whole word. Encoding a single 10 000-char word took 20.6 s on net8.0 before and takes about 0.35 ms now. This matters if you tokenize untrusted input. Output is unchanged.
• LoadFromHuggingFaceAsync on netstandard2.0 (#​159), so it is also available on .NET Framework.

Deprecations

• Decode(ReadOnlySpan<long> tokenIds, bool cleanupTokenizationSpaces) is obsolete (#​107, #​159). Use Decode(ReadOnlySpan<long> tokenIds), which takes the value from decoder.cleanup in tokenizer.json (default true; always true for vocabularies loaded with LoadVocabulary).
  • Compiled binaries keep working unchanged. When you recompile, a call like Decode(ids) switches to the new overload. Output only differs if your tokenizer.json sets "cleanup": false.

Behaviour changes

• tokenizer.json files with added_tokens tokenize differently. Earlier versions ignored added_tokens; the resulting token ids now match Hugging Face. Configurations with an added token that has single_word: true are rejected with an ArgumentException.
• Decode of a sequence starting with a suffix token now puts the configured continuing-subword prefix (e.g. ##) in front of the result ("##m ipsum" instead of "m ipsum").
• The net6.0 target was removed. .NET 6 and 7 apps now use the netstandard2.0 build. Its public API is identical, except for the [Experimental] CreateAsyncBatchEnumerator(ChannelReader<…>, …) overload, which needs net8.0 or later.
• On .NET Framework, LoadFromHuggingFaceAsync needs TLS 1.2. Apps targeting .NET Framework below 4.7 may have to set ServicePointManager.SecurityProtocol themselves.

Other

• Verifiable releases (#​160). The package is built and published to nuget.org by GitHub Actions, using trusted publishing. Each release comes with a build provenance attestation for the .nupkg and for the FastBertTokenizer.dll files inside it. nuget.org adds its own signature to the package, which changes the .nupkg's hash. To verify a package from nuget.org, either:

  • use NuGet Provenance Verifier, which removes the nuget.org signature and checks the package against GitHub's attestations. It runs in your browser and takes a package id and version or a .nupkg file, or
  • extract an assembly from the package and run

    gh attestation verify FastBertTokenizer.dll -R georg-jung/FastBertTokenizer

  GitHub releases are immutable from this version on; gh release verify v1.2.6 -R georg-jung/FastBertTokenizer checks the release itself.

• API docs are published at https://fastberttokenizer.gjung.com/, which now also links to these release notes.

• Package versions follow SemVer 2. The package is validated against 1.0.28 to catch accidental API breaks.

• Dependency floors for netstandard2.0: System.Memory 4.6.3 (was 4.5.5), System.Text.Json 8.0.6 (was 8.0.3).

• Many new parity tests against Hugging Face tokenizers. Known remaining divergences are documented as skipped tests (#​151).

• Reworked benchmarks (.NET 10, fixed baselines, current competitors) with fresh results in the README (#​123, #​158).

Full Changelog: v1.0.28...v1.2.6

1.1.30-alpha

• Add ## (or what is configured in tokenizer.json) to the beginning of the decode result if the first encoded token id is a suffix token.

1.1.21-alpha

Full Changelog: v1.0.28...v1.1.21-alpha

• Add support for added_tokens, #​100
• Add proper support for strip_accents
  • todo: add tests for this

Commits viewable in compare view.

Updated Microsoft.ML.OnnxRuntime from 1.30.0 to 1.31.0.

Release notes

Sourced from Microsoft.ML.OnnxRuntime's releases.

1.31.0

ONNX Runtime 1.31.0 stabilizes model-package and EPContext data APIs, improves CPU model loading and quantized MoE inference, expands device-based execution-provider selection, and strengthens model-loading and runtime reliability. These notes cover changes since ONNX Runtime 1.30.0. CUDA and WebGPU kernel updates are summarized here; detailed provider notes are covered by their separate plugin EP releases.

Highlights

• Promoted the model-package API and EPContext data callbacks to stable C and C++ APIs, with EPContext callback support added across language bindings (#​33166, #​32265).
• Reduced CPU session-initialization overhead by parallelizing eligible weight prepacking, and accelerated block-wise INT4/INT8 QMoE experts with MLAS QNBit kernels and grouped expert dispatch (#​31691, #​32644, #​32668).
• Added x86 FP16 LayerNorm/RMSNorm acceleration, Arm KleidiAI SVE2.1 FP16 GEMM support, and additional RISC-V vector kernels (#​32715, #​32670, #​32710).
• Enabled CoreML participation in device-based EP selection, including Apple Neural Engine selection through PREFER_NPU (#​31975).
• Added workspace-memory accounting and verification for constrained-memory graph partitioning, alongside broader model, graph, and tensor validation (#​31962, #​32189).

Announcements & Compatibility

• CUDA Plugin EP v0.3.0 and WebGPU Plugin EP v0.5.0 are matched with ONNX Runtime 1.31.0. For contrib-op compatibility, core and the plugin must use the same contributed-operator schemas; building both from the same ONNX Runtime revision is recommended. Schema mismatches can cause incorrect execution or crashes and may not be detected during plugin registration.
• GatherBlockQuantized behavior change: out-of-range indices now produce zero output slices across CPU, CUDA, and WebGPU for both integer and floating-point quantized formats. This deliberately differs from ONNX Gather error semantics (#​32480).
• Experimental API migration: model-package consumers should use OrtApi::GetModelPackageApi and the stable OrtModelPackageApi table. EPContext callback registration now uses stable APIs; the corresponding experimental lookup entries were removed. The stable native callback contract carries callback/state pointers, with payload limits enforced by application callbacks, bindings, or EPs rather than a native read-options policy (#​33166, #​32265).
• ACL EP is deprecated and will be removed in a future release. Source builds using --use_acl now receive a deprecation warning (#​32564).
• The ONNX dependency remains at 1.22.0. The initial ONNX 1.23 integration was reverted pending upstream fixes; this release does not ship that integration's new opset or FLOAT6 support (#​33073).
• Supported native builds now default to 1DS telemetry. Windows TraceLogging remains selectable with --use_windows_telemetry or onnxruntime_USE_WINDOWS_TELEMETRY=ON; telemetry can be disabled with --no_telemetry (#​32384).

New Features

Core APIs & Runtime

• Stabilized the model-package C and C++ APIs for loading packages, selecting compatible model variants, and creating sessions (#​33166).
• Stabilized named EPContext data read/write callbacks for application-managed storage. Added EP support negotiation so registered callbacks do not silently fall back to filesystem I/O (#​32265).
• Extended the Compile API to write external initializers and EP context data to buffers, enabling in-memory compilation workflows for models larger than 2 GB (#​31347).
• Added workspace-memory accounting, source reporting, and post-partition reservation verification. Set session.strict_workspace_verification=1 to fail initialization on workspace declaration overruns or nonzero orphaned reservations; strict verification is not supported for ORT-format model loads (#​31962, #​32189).
• Added the opt-in source-build option onnxruntime_DISABLE_DEVICE_DISCOVERY=ON to skip platform accelerator probing and report only the CPU device (#​32592).

Shared Contrib Operators

These updates affect core operator schemas and cross-provider behavior, not only the separately released GPU plugins.

• Extended EngramGate and NGramHashMapping for Qwen4-Exp, and added VarlenNGramHashMapping for packed DeepSeek Engram batching with request-local history. Packed hashing also supports Qwen4-Exp head offsets, EOS resets, and segment boundaries (#​32285, #​32358, #​32467).
• Added the activation attribute to GatedRMSNorm for SiLU, its Swish alias, and Sigmoid gating across CPU, CUDA, and WebGPU. SiLU remains the default (#​32512).
• Added weightless hyper-connection operators BranchwiseRMSNorm, ScaledSiLU, HyperConnectionPreMix, and HyperConnectionPostMix, with CPU, CUDA, and WebGPU implementations (#​32687).
• Extended GatherBlockQuantized to FP8/FP4 data, full-axis blocks, and scale broadcasting, and improved CPU UINT8 dequantization. See the out-of-range-index behavior change above (#​32480, #​32621).

Plugin Execution Provider Infrastructure

• Added support for statically linked plugin EPs, with a static WebGPU plugin build configuration (#​32395).
• Preserved caller streams in single-tensor plugin data transfers and handled null plugin allocator returns (#​32666, #​32683).
• Deduplicated AutoEP custom-op domains and corrected EP assignment reporting after NHWC layout transformation (#​32711, #​33184).
• Added reference guidance for compiled-model compatibility validation, including multi-device scenarios (#​29168).

Execution Provider Updates

CUDA EP

CUDA updates include sparse attention and packed continuous batching, expanded paged speculative decoding, INT2 and mixed-width QMoE execution, opt-in MatMul auto-tuning, improved MXFP4 support on Blackwell consumer GPUs, and reliability fixes.
... (truncated)

Commits viewable in compare view.

Updated Microsoft.ML.Tokenizers from 2.0.0 to 2.0.1.

Release notes

Sourced from Microsoft.ML.Tokenizers's releases.

2.0.1-Preview

Minor update for 2.0.0 that introduces a new api for ProduceWordBag so you can pass in a column that already has the weights set.

Commits viewable in compare view.

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore <dependency name> major version will close this group update PR and stop Dependabot creating any more for the specific dependency's major version (unless you unignore this specific dependency's major version or upgrade to it yourself)
• @dependabot ignore <dependency name> minor version will close this group update PR and stop Dependabot creating any more for the specific dependency's minor version (unless you unignore this specific dependency's minor version or upgrade to it yourself)
• @dependabot ignore <dependency name> will close this group update PR and stop Dependabot creating any more for the specific dependency (unless you unignore this specific dependency or upgrade to it yourself)
• @dependabot unignore <dependency name> will remove all of the ignore conditions of the specified dependency
• @dependabot unignore <dependency name> <ignore condition> will remove the ignore condition of the specified dependency and ignore conditions