Skip to content

Pull requests: NVIDIA/Model-Optimizer

New pull request New
276 Open 1,532 Closed
276 Open 1,532 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

Validate Puzzletron campaign configuration before launch puzzletron_v2 Related to feature/puzzletron_v2 branch
#2249 opened Aug 25, 2026 by j-rausch Contributor Loading…
9
fix(quantization): disable the KV cache for layerwise calibration
#2248 opened Aug 25, 2026 by Fridah-nv Contributor Loading…
51
specdec_bench: emit speculation_profile.json alongside acceptance metrics
#2247 opened Aug 25, 2026 by yeyu-nvidia Contributor Loading…
11
Add Aumann-Shapley AutoQuantize recipe integration
#2246 opened Aug 25, 2026 by joshua-hill Loading…
35
Automate Puzzletron setup and fix unattended runtime bugs
#2245 opened Aug 25, 2026 by j-rausch Contributor Loading…
16
Add NVFP4/FP8 Q/K/P/V + 2:4 attention quantization for MLA (vLLM TRITON_MLA)
#2244 opened Aug 25, 2026 by kaix-nv Contributor • Draft
5
[6648996] Fix NVFP4 ONNX packed-weight scale rounding
#2243 opened Aug 25, 2026 by ajrasane Contributor • Draft
4
Update documentation theme to Shibuya
#2242 opened Aug 24, 2026 by realAsma Contributor Loading…
4
[ONNX][Autocast] Adds nodes_to_exclude regex support to the QDQ-aware convert_to_f16() API
#2241 opened Aug 24, 2026 by jai17 Loading…
8
[ONNX] Add quantization sensitivity ranking + exclusion picker
#2240 opened Aug 24, 2026 by gcunhase Contributor • Draft
10
[4/n] Add joint mask reuse topology selection
#2237 opened Aug 24, 2026 by kaix-nv Contributor • 4/4 • Draft
4
[3/n] Add vLLM mask reuse capture
#2236 opened Aug 24, 2026 by kaix-nv Contributor • 3/4 • Draft
4
[2/n] Add target-first offline mask reuse calibration
#2235 opened Aug 24, 2026 by kaix-nv Contributor • 2/4 • Draft
4
[chore]: weekly bump of uv.lock on main (2026-08-24)
#2234 opened Aug 24, 2026 by github-actions Bot Loading…
2
fix(launcher): use the schema's draft_model global var, and validate global_vars keys
#2232 opened Aug 24, 2026 by h-guo18 Contributor Loading…
6
Fix distributed AutoQuantize scoring and share backward setup
#2231 opened Aug 23, 2026 by joshua-hill Loading…
24
Fix Diffusers export dtype resolution
#2225 opened Aug 20, 2026 by realAsma Contributor Loading…
21
Set NVFP4 MSE preset effective bits to 4.5
#2223 opened Aug 20, 2026 by realAsma Contributor Loading…
2
fix(speculative): read rope_theta from rope_parameters first
#2221 opened Aug 20, 2026 by h-guo18 Contributor Loading…
6
Restructure recipes: split per-model_type recipes from model-hub checkpoint recipes
#2219 opened Aug 20, 2026 by shengliangxu Collaborator Loading…
22
Kimi-K3 on layerwise fused export: 3 defect fixes, multimodal support, single-B200 NVFP4
#2218 opened Aug 19, 2026 by Fridah-nv Contributor • 2/2 • Draft
4
[Speculative Decoding] DFlash2 draft variant (sublayer convolution + candidate selector)
#2216 opened Aug 19, 2026 by h-guo18 Contributor Loading…
22
Group MLA q_a_proj/kv_a_proj_with_mqa projections in auto_quantize
#2213 opened Aug 18, 2026 by joshua-hill • Draft
2
Add layer-wise KV-cache AutoQuant with forward KL
#2211 opened Aug 18, 2026 by meenchen Contributor • Draft
@meenchen
12
[OMNIML-5774] Add BEVFormer ONNX PTQ and evaluation example
#2208 opened Aug 18, 2026 by ajrasane Contributor • Draft
2
ProTip! Updated in the last three days: updated:>2026-08-22.