-
Notifications
You must be signed in to change notification settings - Fork 350
Pull requests: microsoft/onnxruntime-genai
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Reuse the block drafter runtime for DSpark
#2497
opened Aug 28, 2026 by
Tianlei Wu (tianleiwu)
Contributor
•
7/7
•
Draft
4 of 5 tasks
Run a DFlash2 block drafter in the batch engine
#2496
opened Aug 28, 2026 by
Tianlei Wu (tianleiwu)
Contributor
•
6/7
•
Draft
5 tasks done
Run chained MTP drafting in the batch engine
#2495
opened Aug 28, 2026 by
Tianlei Wu (tianleiwu)
Contributor
•
5/7
•
Draft
4 tasks done
Configure MTP models for paged runtime decoding
#2494
opened Aug 28, 2026 by
Tianlei Wu (tianleiwu)
Contributor
•
4/7
•
Draft
Expose speculative draft proposals through runtime APIs
#2493
opened Aug 28, 2026 by
Tianlei Wu (tianleiwu)
Contributor
•
3/7
•
Draft
Add support for quant_auto format in model builder
#2492
opened Aug 28, 2026 by
Vishal Jain (VishalX)
Contributor
Loading…
Verify speculative drafts in dynamic Engine steps
#2491
opened Aug 28, 2026 by
Tianlei Wu (tianleiwu)
Contributor
•
2/7
•
Draft
Export DSpark block drafters for Qwen 3.5 and 3.8
#2490
opened Aug 28, 2026 by
Tianlei Wu (tianleiwu)
Contributor
Loading…
Allow top_k greater than vocab_size
#2486
opened Aug 28, 2026 by
Hrishith Thadicherla (hthadicherla)
Contributor
Loading…
Generalize Gemma 4 MoE Quark path to plain Quark/AWQ int4 experts
#2483
opened Aug 27, 2026 by
Thiago Pereira Rocha (thpereir)
Contributor
Loading…
3 tasks done
Add 2-bit (uint2) Gemma 4 MoE support to the model builder
#2482
opened Aug 27, 2026 by
Thiago Pereira Rocha (thpereir)
Contributor
Loading…
3 tasks done
adds engine benchmark to integration tests
#2481
opened Aug 27, 2026 by
Prathik Rao (prathikr)
Contributor
Loading…
[C#] Expose cached Generator through IChatClient.GetService
#2475
opened Aug 27, 2026 by
ZedingZhang (ZedingZhang)
Loading…
Add Gemma 4 text model builder support (dense gemma-4-12b-it + 26B-A4B MoE)
#2473
opened Aug 26, 2026 by
Thiago Pereira Rocha (thpereir)
Contributor
Loading…
[Engine] Add paged hybrid state and speculative drafters
#2472
opened Aug 26, 2026 by
Tianlei Wu (tianleiwu)
Contributor
•
Draft
Support stateful KV cache for multimodal decoder
#2469
opened Aug 25, 2026 by
Anirudh Swaminathan (Anirudh-Swaminathan)
•
Draft
[benchmark] Add prompt token truncation and trim benchmark output
#2468
opened Aug 25, 2026 by
Namal Rajatheva (amd-namalr)
Loading…
Optimize Engine guidance for production workloads
#2466
opened Aug 25, 2026 by
Baiju Meswani (baijumeswani)
Collaborator
Loading…
Redesign experimental Engine turn API
#2464
opened Aug 24, 2026 by
bmehta001
Contributor
Loading…
1 task
Commit accepted speculative prefixes in fixed state pools
#2455
opened Aug 23, 2026 by
Tianlei Wu (tianleiwu)
Contributor
•
Draft
Replay compact fixed state for hybrid decoding
#2454
opened Aug 23, 2026 by
Tianlei Wu (tianleiwu)
Contributor
•
1/7
Loading…
4 tasks done
Keep Python opaque data alive for the lifetime of the request
#2445
opened Aug 21, 2026 by
Gopalakrishnan Nallasamy (GopalakrishnanN)
Contributor
Loading…
Previous Next
ProTip!
Follow long discussions with comments:>50.