Skip to content

Pull requests: NVIDIA/Model-Optimizer

New pull request New
174 Open 1,121 Closed
174 Open 1,121 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[Example] Add 2:4 sparsity -> INT8 SmoothQuant PTQ -> ONNX -> TensorRT pipeline
#1664 opened Jun 10, 2026 by ajrasane Contributor • Draft
2
[OMNIML-5024] specdec_bench cell t0_d3 — google/gemma-4-E4B-it / MTP / vllm
#1663 opened Jun 10, 2026 by ChenhanYu Collaborator • Draft
2
docs: add modelopt_recipes README and PTQ recipe/scheme guide
#1662 opened Jun 9, 2026 by cjluo-nv Collaborator Loading…
11
[6241485] Add support for ONNX Q/DQ node placement for DLA
#1661 opened Jun 9, 2026 by gcunhase Contributor Loading…
2
Add NVFP4 + QAD to the Nemotron-3-Nano-30B-A3B tutorial
#1660 opened Jun 9, 2026 by kevalmorabia97 Collaborator • Draft
3
Add fused Triton kernel for local-Hessian NVFP4 weight-scale search
#1659 opened Jun 9, 2026 by Fridah-nv Contributor Loading…
17
docs(deployment skill): drop wrong "release predates arch" cu130 fallback
#1654 opened Jun 8, 2026 by Edwardf0t1 Contributor Loading…
5
feat(deepseek): add --cast_mxfp4_to_nvfp4 to deepseek_v4 quantize step cherry-pick-0.45.0 After code freeze, cherry-pick to release branch for next rc (bulk update). Only for bug fixes / doc
#1653 opened Jun 8, 2026 by cjluo-nv Collaborator Loading…
14
docs(eval skill): drop arch-specific cu130 nightly tag (release images are multi-arch)
#1649 opened Jun 8, 2026 by cjluo-nv Collaborator Loading…
4 of 5 tasks
2
[OMNIML-4944] peft: add lora_dtype field to PEFTAttributeConfig
#1646 opened Jun 8, 2026 by hychiang-git Contributor • Draft
3
feat: Layerwise calibration memory optimizations (non-mutating skip + meta placeholders)
#1640 opened Jun 5, 2026 by Fridah-nv Contributor • Draft
2
Support MCore auto_quantize calibration updates
#1639 opened Jun 5, 2026 by realAsma Contributor • Draft
2
[OMNIML-4962] specdec_bench cell t0_d3 — Qwen/Qwen3.5-4B / DFlash / vLLM
#1638 opened Jun 5, 2026 by ChenhanYu Collaborator Loading…
10
Add NVFP4 fakequant for attention BMMs
#1635 opened Jun 5, 2026 by kaix-nv Contributor • Draft
2
Route decode-only skip-softmax through the decode kernel in vLLM
#1634 opened Jun 5, 2026 by kaix-nv Contributor • Draft
2
fix(export): correct unified_export_megatron at EP > 1 and DP > 1
#1631 opened Jun 4, 2026 by yueshen2016 Contributor Loading…
3 of 4 tasks
9
remove deprecated get_default_load_sharded_strategy
#1629 opened Jun 4, 2026 by dimapihtar Loading…
4
[6058841] Consistent types on If/Loop/Scan subgraphs during FP16/BF16 conversion cherry-pick-0.45.0 After code freeze, cherry-pick to release branch for next rc (bulk update). Only for bug fixes / doc
#1628 opened Jun 4, 2026 by ajrasane Contributor • Draft
10
[6058907] Fix ShapeInferenceError in ONNX int8+fp16 quantization of weakly-typed models cherry-pick-0.45.0 After code freeze, cherry-pick to release branch for next rc (bulk update). Only for bug fixes / doc
#1627 opened Jun 4, 2026 by ajrasane Contributor Loading…
9
Add efficient Triton decode attention kernel with fused skip-softmax …
#1624 opened Jun 3, 2026 by kaix-nv Contributor • Draft
4
Skip-Softmax calibration in vLLM
#1622 opened Jun 3, 2026 by kaix-nv Contributor • Draft
2
DFlash speculative decoding for MiniMax-M2.7 (FSDP2): auto mask-token, FSDP2 resume fixes, per-checkpoint draft export
#1621 opened Jun 3, 2026 by yeyu-nvidia Contributor Loading…
10
Add W4A16 NVFP4-MSE Qwen3.5 dense/MoE PTQ recipes
#1620 opened Jun 3, 2026 by cjluo-nv Collaborator Loading…
2
[Feat]: Specdec Streaming: RDMA + Multinode
#1611 opened Jun 2, 2026 by h-guo18 Contributor Loading…
@h-guo18
27
Fix torch import error to remove circular dependency & move Nemotron configs
#1606 opened Jun 2, 2026 by jenchen13 Contributor Loading…
10
ProTip! Add no:assignee to see everything that’s not assigned.