fix(deps, pyamber): bump transformers from 5.3.0 to 5.10.1 in /amber

[dependabot]

Bumps transformers from 5.3.0 to 5.10.1.

Release notes

Sourced from transformers's releases.

Release v5.10.1

v5.10.0 was yanked as we publish on a corrupted branch. Sorry everyone, this happens when we rush a release!!!

New Model additions

Gemma4 unified+ Gemma4 MTP

Gemma 4 12B Unified is an encoder-free multimodal model with pretrained and instruction-tuned variants. Unlike standard Gemma 4, which uses dedicated encoder towers, Gemma 4 12B Unified projects raw inputs directly into the language model's embedding space through lightweight linear pipelines. This results in a simpler architecture while maintaining strong multimodal performance.

Key differences from standard Gemma 4:

• No Vision Tower: Raw pixel patches are projected directly into LM space via a Dense + LayerNorm pipeline with factorized 2D positional embeddings, replacing the vision encoder.
• No Audio Tower: Raw 16 kHz waveform samples are chunked into fixed-length frames and projected through a simple RMSNorm → Linear pipeline, replacing the mel spectrogram + Conformer encoder.
• Shared Multimodal Pipeline: Both vision and audio use the same Gemma4UnifiedMultimodalEmbedder (RMSNorm → Linear) for the final projection to text hidden space.

You can find the original Gemma 4 12B Unified checkpoints under the Gemma 4 release.

• who needs encoders? (#46385) by @​douglas-reid @​sgerrard @​vasqu @​molbap

Sapiens2

Sapiens2 is a family of high-resolution vision transformers pretrained on ~1 billion curated human images, designed for human-centric computer vision tasks including pose estimation, body-part segmentation, surface normal estimation, and pointmap estimation. The models scale from 0.4B to 5B parameters and train at native 1K resolution, with hierarchical 4K variants for extended spatial reasoning. Sapiens2 achieves substantial improvements over its predecessor with +4 mAP in pose estimation, +24.3 mIoU in body-part segmentation, and 45.6% error reduction in normal estimation.

Links: Documentation | Paper

• Add Sapiens2 Model (#45919) by @​guarin in #45919

DeepSeek-OCR-2

DeepSeek-OCR-2 is an OCR-specialized vision-language model built on a distinctive architecture that combines a SAM ViT-B vision encoder with a Qwen2 hybrid attention encoder, connected through an MLP projector to a DeepSeek-V2 Mixture-of-Experts (MoE) language model. The model features a hybrid attention mechanism that applies bidirectional attention over image tokens and causal attention over query tokens, enabling efficient and accurate document understanding. It supports both plain OCR tasks and grounding capabilities with coordinate-aware output for document conversion to markdown format.

Links: Documentation

• Add Deepseek-OCR-2 model (#45075) by @​thisisiron in #45075

Mellum

Mellum is a code-focused Mixture-of-Experts language model developed by JetBrains. It is derived from the Qwen3-MoE architecture with per-layer-type RoPE and interleaved sliding window attention. The model has 12B total parameters with 2.5B active parameters per token, using 64 routed experts with 8 activated per token across 28 layers.

Links: Documentation

• feat: Add support for JetBrains' Mellum v2 code generation model (#46112) by @​shadeMe in #46112

Breaking changes

The Gemma4 vision pooler now casts inputs to float32 before scaling to prevent float16 overflow (inf saturation) with large checkpoints, which may cause minor numerical differences in outputs for users running Gemma-4 vision models in float16.

• 🚨 Fix float16 overflow in Gemma4 vision pooler (#46277) by @​Bluear7878

Audio Language Models (ALMs) now have a dedicated base model class without a language modeling head, aligning them with the design of Vision Language Models (VLMs); users relying on the previous model class structure should update their code to use the new base model class where appropriate.

• 🚨 [ALM] Add base model without head (#45534) by @​eustlb

... (truncated)

Commits

• 90c3ae5 Patch because we had to yank 5.10 because the release branch was not up to date
• 0bd94b3 v5.10.0
• 1423d22 who needs encoders? (#46385)
• 50eb20a Fix dsv4 dequant + tp/ep (#46378)
• 74464e8 Fix wrong changes produced by style/repo. check bot (#46371)
• 1b8ec34 Fix path traversal when saving Bark voice preset embeddings (#46237)
• e820678 Add Sapiens2 Model (#45919)
• 595721c Pass library_name/version to Hub calls via a shared HfApi (#46318)
• 0f0036c docs: update ACL Anthology URL in CITATION.cff (#46352)
• fa6c830 DeepGEMM BF16 + mixed FP8/FP4 + MegaMoE + refactor (#45634)
• Additional 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 this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)
  You can disable automated security fix PRs for this repo from the Security Alerts page.

[github-actions]

Automated Reviewer Suggestions

Based on the git blame history of the changed files, we recommend the following reviewers:

• Contributors with relevant context: @aglinxinyuan
  You can notify them by mentioning @aglinxinyuan in a comment.

[github-actions]

⚠️ Benchmark changes need a look

🟢 2 better · 🔴 3 worse · ⚪ 10 noise (<±5%) · 0 without baseline

Compared against main e029e0d benchmarked on this same runner, so the delta is largely free of cross-runner hardware noise. The "7d avg" column still reflects the gh-pages dashboard. Treat <±5% as noise unless repeated.

Dashboard · Run

	config	throughput	MB/s	latency	max Δ latest / 7d
🔴	bs=10 sw=10 sl=64	385	0.235	25,605/32,979/32,979 us	🟢 -11.6% / ⚪ within ±5%
🔴	bs=100 sw=10 sl=64	809	0.494	123,887/141,435/141,435 us	🔴 +5.8% / ⚪ within ±5%
⚪	bs=1000 sw=10 sl=64	925	0.565	1,086,152/1,116,676/1,116,676 us	⚪ within ±5% / ⚪ within ±5%

Baseline details

Latest main e029e0d from same runner

config	metric	PR	latest main	7d avg	Δ latest	Δ 7d
bs=10 sw=10 sl=64	throughput	385 tuples/sec	392 tuples/sec	n/a	-1.8%	n/a
bs=10 sw=10 sl=64	MB/s	0.235 MB/s	0.239 MB/s	n/a	-1.7%	n/a
bs=10 sw=10 sl=64	p50	25,605 us	24,358 us	n/a	+5.1%	n/a
bs=10 sw=10 sl=64	p95	32,979 us	37,288 us	n/a	-11.6%	n/a
bs=10 sw=10 sl=64	p99	32,979 us	37,288 us	n/a	-11.6%	n/a
bs=100 sw=10 sl=64	throughput	809 tuples/sec	835 tuples/sec	n/a	-3.1%	n/a
bs=100 sw=10 sl=64	MB/s	0.494 MB/s	0.51 MB/s	n/a	-3.1%	n/a
bs=100 sw=10 sl=64	p50	123,887 us	118,893 us	n/a	+4.2%	n/a
bs=100 sw=10 sl=64	p95	141,435 us	133,730 us	n/a	+5.8%	n/a
bs=100 sw=10 sl=64	p99	141,435 us	133,730 us	n/a	+5.8%	n/a
bs=1000 sw=10 sl=64	throughput	925 tuples/sec	916 tuples/sec	n/a	+1.0%	n/a
bs=1000 sw=10 sl=64	MB/s	0.565 MB/s	0.559 MB/s	n/a	+1.1%	n/a
bs=1000 sw=10 sl=64	p50	1,086,152 us	1,087,178 us	n/a	-0.1%	n/a
bs=1000 sw=10 sl=64	p95	1,116,676 us	1,157,945 us	n/a	-3.6%	n/a
bs=1000 sw=10 sl=64	p99	1,116,676 us	1,157,945 us	n/a	-3.6%	n/a

Raw CSV

config_idx,batch_size,schema_width,string_len,num_batches,total_ms,total_tuples,total_bytes,tuples_per_sec,mb_per_sec,lat_p50_us,lat_p95_us,lat_p99_us
0,10,10,64,20,519.77,200,128000,385,0.235,25604.81,32979.46,32979.46
1,100,10,64,20,2471.78,2000,1280000,809,0.494,123886.70,141435.05,141435.05
2,1000,10,64,20,21619.64,20000,12800000,925,0.565,1086151.77,1116675.60,1116675.60