NVIDIA Nemotron 0.2.0

What's Changed

• Added Ultra README updates by @anniesurla in #118
• docs: Add 0.1.0 docs version to picker by @chtruong814 in #121
• docs: Swap our super image by @ko3n1g in #128
• Add Spark Deployment Guide by @chrisalexiuk-nvidia in #129
• Nemotron cc docs by @ayushdg in #95
• Update Spark Configs by @chrisalexiuk-nvidia in #131
• Add embedding recipe: build domain-specific embeddings from raw documents by @oliverholworthy in #85
• Remove Nemotron 3 Super Advanced Deployment Guide by @chrisalexiuk-nvidia in #133
• ci: Add claude review to nemotron by @chtruong814 in #138
• Update vLLM to pinned version 0.18.1 by @chrisalexiuk-nvidia in #135
• Pin vllm version by @chrisalexiuk-nvidia in #140
• Expand embedding recipe README with comprehensive documentation by @shan-nvidia in #134
• fix(super3): add missing data subcommand package by @Doondi-Ashlesh in #145
• Fix retriever SDG dedup plugin circular import by @oliverholworthy in #156
• Fix embedding SDG dedup compatibility. by @hoangnguyenN94 in #147
• docs: add SECURITY.md by @chtruong814 in #160
• fix(super3): point SFT HF conversion at Nemotron-3-Super base by @shashank3959 in #158
• fix(embed): propagate query/passage_max_length to hard negative mining by @shan-nvidia in #163
• Add Nemotron 3 Nano Omni Document Intelligence cookbook by @chiachihchen in #169
• Add Nemotron-3-Nano Omni cookbooks for vLLM, SGLang, and TensorRT-LLM by @anushapant in #172
• docs(automodel): add Nemotron-Nano-Omni V3 fine-tuning cookbook by @HuiyingLi in #170
• Nemotron Nano Omni V3 RL assets by @vinhngx in #168
• Nemotron 3 Nano Omni SFT cookbook by @vinhngx in #171
• Adding Omni-3 recipes by @marcromeyn in #173
• Update TRT-LLM setup for Nemotron-3-Nano-Omni by @anushapant in #175
• Add support for Lepton and DGX Cloud executors by @rapaul-nv in #162
• Update Nemotron 3 Nano Omni deployment cookbooks by @anushapant in #181
• Add overviews and tables of contents to Nemotron-3-Nano-Omni cookbooks by @anushapant in #187
• docs: Replace READMEs with summary pages by @mikemckiernan in #177
• docs(automodel): add Step 4/5 + baseline inference + live loss viz to Nemotron-Omni cookbook by @HuiyingLi in #188
• feat(embed): multi-GPU finetune via shared nemo_runspec.execute_uv_local by @oliverholworthy in #165
• Bulk updates for the 1.2.0 release by @ayushdg in #205
• Update Nemotron 3 Nano Omni vLLM deployment cookbook by @anushapant in #201
• Playbook rc1 by @rkalaniNV in #198
• Create team-request.yml by @rkalaniNV in #217
• fix(docs): Replace broken link for mmlongbench-doc by @mikemckiernan in #178
• Fix data SDG optional imports at CLI startup by @oliverholworthy in #210
• Add optimizer backend selection to embed finetune recipe by @oliverholworthy in #211
• Refactor team-request workflow for skill onboarding by @rapaul-nv in #223
• docs: Refactor model recipe presentation by @mikemckiernan in #185
• docs: Refactoring prep for steps by @mikemckiernan in #221
• docs: Add SDG steps documentation by @mikemckiernan in #222
• docs: Add translation step documentation by @mikemckiernan in #224
• docs: Add model training w/ steps by @mikemckiernan in #225
• docs: Add MCQ benchmark step by @mikemckiernan in #226
• docs: Add model eval step by @mikemckiernan in #227
• docs: Add step basics by @mikemckiernan in #228
• docs: Add data curation step by @mikemckiernan in #229
• docs: Wrap up steps docs by @mikemckiernan in #231
• Enhance nemotron-customize documentation and developer journeys by @rapaul-nv in #233
• Update author metadata in SKILL.md for nemotron-customize by @rapaul-nv in #234
• feat(rerank): add cross-encoder reranking recipe by @oliverholworthy in #139
• Super v3 cookbook revision by @vinhngx in #206
• Update nemotron-customize SKILL.md by @rapaul-nv in #236
• Add retrieval recipes agent skill by @oliverholworthy in #235
• docs(fix): Fix broken links by @mikemckiernan in #241
• docs(fix): Redirects for cookbooks and use cases by @mikemckiernan in #242
• chore: No SEO for nightly by @mikemckiernan in #243
• docs(fix): Fix TOC for training recipes by @mikemckiernan in #232
• Add computer use agent Omni cookbook by @anjalibshah in #191
• Distributed multi-GPU conversion, plus data-prep/sharding and config/checkpoint fixes by @rapaul-nv in #238
• Add Nemotron 3.5 Content Safety usage cookbooks by @svenchilton in #247
• Add Nemotron Policy Generator skill by @svenchilton in #237
• Translation improvements by @rkalaniNV in #249
• adding Ultra v3 RL, automodel and Mbridge cookbooks by @vinhngx in #251
• Add Nemotron 3 Ultra deployment cookbooks with vLLM, SGLang, and TensorRT-LLM by @anushapant in #250
• Adding Nemotron 3 Ultra by @marcromeyn in #252
• Ultra - Agentic Resources by @chrisalexiuk-nvidia in #253
• Update vLLM NVFP4 serving config by @anushapant in #254
• Trigger nvskills ci by @rkalaniNV in #255
• Add NIM deployment cookbook and README index for Nemotron 3.5 Content Safety by @svenchilton in #256
• docs(ultra3): update quantization recipe and precision table by @yueshen2016 in #257
• Add Spark Deployment Guide for Ultra by @chrisalexiuk-nvidia in #262
• Add private OCI OKE cookbook for Llama Nemotron Nano 8B by @fede-kamel in #117
• docs(oci): keep Nano-8B private-OKE cookbook working on current OKE (follow-up to #117) by @fede-kamel in #267
• ci: use NVIDIA inference for Claude review by @chtruong814 in #282
• Add CODE_OF_CONDUCT.md and GitHub issue/PR templates by @M-H-Jishan in #285
• DGX Station Guide for Ultra by @chrisalexiuk-nvidia in #286
• docs: fix broken examples link in CONTRIBUTING.md by @andrewwhitecdw in #277
• docs: fix typos in README and training docs by @andrewwhitecdw in #276
• Fix test imports broken by the kit to nemo_runspec refactor by @temujinkz in #275
• Fix packer dropping rows on null or whitespace-only reasoning_content (Fixes #240) by @temujinkz in #274
• docs: fix broken nemo_runspec package-readme links by @temujinkz in #268
• Apply per-shard row partition in binidx pretrain processing by @temujinkz in #272
• Fix local data prep path doubling in filesystem discovery by @temujinkz in #273
• feat/sdg bash notebook by @chrisalexiuk-nvidia in #62
• Fix qrel source identifier collisions by @shan-nvidia in #287
• Support JSONL files in embed prep directories by @shan-nvidia in #288
• Add improved deployment config for Nemotron 3 Ultra on DGX Station by @chrisalexiuk-nvidia in #292
• Update Station Deployment Guide for better agentic throughput by @chrisalexiuk-nvidia in #293
• feat(embed): update default embedding profile by @oliverholworthy in #291
• Use Nemotron 3 embed for dedup embeddings by @shan-nvidia in #297
• chore(ci): AUT-1178 pin GitHub Actions to commit SHAs by @svcnemo-autobot in #305
• chore: Sep PR for Fern CI by @mikemckiernan in #310
• chore(ci): AUT-1231 bump preflight workflow template by @svcnemo-autobot in #311
• fix(docs): pin DGX Station deployment to GB300 by @chrisalexiuk-nvidia in #312
• Nemotron 3.5 Lightning RL cookbook by @vinhngx in #315
• Add Nemotron-3.5 Lightning Text2SQL LoRA recipe (Megatron-Bridge) by @Maghoumi in #316
• Add Nemotron 3.5 Lightning deployment cookbooks with vLLM, SGLang, and TensorRT-LLM by @anushapant in #317
• docs: add Nemotron 3.5 Lightning agentic tools guide by @chrisalexiuk-nvidia in #318
• docs: add Nemotron 3.5 Lightning cookbook index by @chrisalexiuk-nvidia in #319
• Adding training recipe for lightning-3.5 by @marcromeyn in #320
• Add Nemotron-3.5-Lightning AutoModel LoRA Text2SQL cookbook by @abodhankar in #321
• Update Nemotron 3.5 Lightning SGLang cookbook by @anushapant in #322
• docs: fix typos and clarify Dual Station Deployment Guide by @bobby-nandigam in #324
• ci: AUT-1501 bump SSO preflight to v1.8.10 by @svcnemo-autobot in #328
• feat(embed): consume released retrieval SDG package by @shan-nvidia in #326
• Update Nemotron 3.5 Lightning Cookbooks by @anushapant in #333
• Add GRPO recipe for Lightning on 2x Stations by @roclark in #336
• Fix Nemotron Lightning 3.5 VDR by @vinhngx in #335
• Add SFT recipe for Lightning on 2x Stations by @roclark in #339
• Add LoRA recipe for Lightning on DGX Station by @roclark in #340
• docs: add public Sports Intelligence Playbooks links to usage cookbook by @sophiahhuang in #343
• refactor(review): migrate legacy reviews to /review by @ko3n1g in #361
• docs: add Nemotron post-training calculator by @yyu22 in #350
• Super 3.5 VL RL cookbook by @vinhngx in #365
• Add Nemotron-3.5 Super VL finance LoRA recipe (Megatron-Bridge) by @Maghoumi in #368
• Add Nemotron 3.5 Super VL deployment cookbooks by @anushapant in #371
• Add Nemotron 3.5 Super VL visual investigation agent cookbook by @chiachihchen in #372
• Update Lightning GRPO recipe on Station by @roclark in #364
• Add Nemotron-3.5 Super VL usage cookbook README by @chris-alexiuk in #373
• Add NeMo AutoModel LoRA finance recipe for Nemotron 3.5 Super VL by @abodhankar in #369
• Adding Super 3.5 by @marcromeyn in #374
• chore: bump version to 0.2.0 by @marcromeyn in #375

New Contributors

• @ko3n1g made their first contribution in #128
• @Doondi-Ashlesh made their first contribution in #145
• @hoangnguyenN94 made their first contribution in #147
• @HuiyingLi made their first contribution in #170
• @mikemckiernan made their first contribution in #177
• @anjalibshah made their first contribution in #191
• @svenchilton made their first contribution in #247
• @yueshen2016 made their first contribution in #257
• @fede-kamel made their first contribution in #117
• @M-H-Jishan made their first contribution in #285
• @andrewwhitecdw made their first contribution in #277
• @temujinkz made their first contribution in #275
• @svcnemo-autobot made their first contribution in #305
• @bobby-nandigam made their first contribution in #324
• @roclark made their first contribution in #336
• @sophiahhuang made their first contribution in #343
• @yyu22 made their first contribution in #350
• @chris-alexiuk made their first contribution in #373

Full Changelog: v0.1.0...v0.2.0