docs(nemo-retriever): 26.07 RC readiness corrections for NeMo Retriever Library

[kheiss-uwzoo]

Description

Fix customer-facing RC documentation issues:

• Live RAG imports (NVBugs 6492587): Import LiteLLMClient and LLMJudge from the supported public API, nemo_retriever.models.llm. The documented nemo_retriever.llm module is not included in current nemo-retriever wheels and fails with ModuleNotFoundError.
• Local GPU install: Align the CUDA 13 override command with the frozen [local] dependency metadata in nemo_retriever/pyproject.toml: torch==2.11.0 and torchvision==0.26.0.
• Nemotron Parse installer wording: Make uv pip install nemo-retriever[nemotron-parse] the primary command in the extraction overview and support matrix, consistent with the UV-first README, while explicitly identifying plain pip install as equivalent for users who do not use UV.
• Remote NIM platform support: Clarify that supported local GPU inference requires Linux and CUDA, while the base package for remote NIM inference can also be installed on Windows x64 and macOS x64/ARM64.
• Statement-style headings: Rename non-FAQ question-style headings to statement titles per NVIDIA Template Library guidance (overview H1, root index H1, LanceDB section, MkDocs nav). FAQ headings stay as questions. Existing heading anchors are preserved.

The branch was rebased onto current main. Earlier NVAIE clarification and obsolete QA-link cleanup commits were dropped because equivalent updates are already present on main. Heading renames were folded in from #2409 (now closed).

Validation

• DORI code-block validation: passed (13 README blocks checked, no issues)
• DORI documentation/link validation: passed (no broken links)
• pre-commit run for changed documentation files: passed
• NRL extraction page-role and link-CTA leakage checks: passed
• README Torch/Torchvision pins match nemo_retriever/pyproject.toml
• Full strict MkDocs build is blocked by the pre-existing missing THIRD_PARTY_LICENSES.md snippet referenced by docs/docs/license.md; the failure is unrelated to this diff

Checklist

• I am familiar with the Contributing Guidelines.
• The documentation is up to date with these changes.
• Changed-file validation passes.

[greptile-apps]

Greptile Summary

This PR corrects five customer-facing documentation issues for the 26.07 RC: it fixes broken LiteLLMClient/LLMJudge import paths (nemo_retriever.llm → nemo_retriever.models.llm), aligns the CUDA 13 torch/torchvision override command with the frozen [local] pins in pyproject.toml (torch==2.11.0, torchvision==0.26.0), and adds clear platform guidance (local GPU requires Linux; base package also installs on Windows x64 and macOS).

• Import fix (NVBugs 6492587): LiteLLMClient and LLMJudge are now imported from nemo_retriever.models.llm, the documented public API that is actually shipped in the wheel; the old nemo_retriever.llm path is absent from the package.
• Version pin alignment: The CUDA override step in README.md now pins both torch==2.11.0 and torchvision==0.26.0, matching nemo_retriever/pyproject.toml's [local] extra exactly.
• Heading & wording cleanup: Question-style H1s are renamed to statement titles (anchors preserved), uv pip install is made the primary install form for nemotron-parse (with plain pip noted as equivalent), and MkDocs !!! admonitions in prerequisites-support-matrix.md are converted to plain blockquotes for DORI toolchain compatibility.

Confidence Score: 5/5

Safe to merge — all changes are documentation-only corrections that fix broken example code and align version pins with the actual package metadata.

Every factual claim in the diff was verified against the repository: the new import paths (nemo_retriever.models.llm) match the module's init.py and all; the torch/torchvision pins (2.11.0/0.26.0) match nemo_retriever/pyproject.toml's [local] extra exactly; and the platform support statements are consistent with the required-environments comment in pyproject.toml. No code logic is touched.

Files Needing Attention: No files require special attention.

Important Files Changed

Filename	Overview
nemo_retriever/README.md	Fixes LLM import paths (nemo_retriever.llm → nemo_retriever.models.llm), corrects torch/torchvision CUDA 13 pins to match pyproject.toml (2.11.0/0.26.0), splits local-GPU vs. remote-NIM install paths, and clarifies platform requirements.
docs/docs/extraction/prerequisites-support-matrix.md	Adds a platform summary note clarifying Linux/CUDA for local GPU vs. Windows/macOS for remote NIM; converts several MkDocs !!! admonitions to plain blockquotes; updates nemotron-parse install example to use uv pip install as the primary form.
docs/docs/extraction/overview.md	Renames H1 from question style to statement style with anchor preserved; rewrites intro sentences for conciseness; updates nemotron-parse note to lead with uv pip install.
docs/docs/index.md	Renames root H1 from 'What is NVIDIA NeMo Retriever?' to 'NVIDIA NeMo Retriever Overview' with backward-compatible anchor; change is purely cosmetic/heading style.
docs/docs/extraction/vdbs.md	Renames 'Why LanceDB?' section heading to 'LanceDB Overview' while preserving the #why-lancedb anchor so existing deep links continue to work.
docs/docs/extraction/audio-video.md	Updates two link labels from 'What is NeMo Retriever Library?' to 'NeMo Retriever Library Overview' to match the renamed H1 in overview.md.
docs/docs/extraction/getting-started-about.md	Updates link label for overview.md and adds inline note that local GPU inference requires Linux while remote NIM also supports Windows/macOS.
docs/docs/extraction/multimodal-extraction.md	Updates four link labels from 'What is NeMo Retriever Library?' to 'NeMo Retriever Library Overview' to reflect the renamed H1.
docs/docs/extraction/troubleshoot.md	Updates one link label for overview.md; troubleshoot pip install example intentionally remains as plain pip (not uv), per the PR's policy of uv as primary only in overview and support matrix.
docs/mkdocs.yml	Updates nav label from 'What is NeMo Retriever?' to 'NeMo Retriever Library Overview' to match the renamed H1 in overview.md.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[User reads README / docs] --> B{Inference path?}
    B -->|Local GPU - Linux only| C["Install: uv pip install nemo-retriever[local]"]
    B -->|Remote NIM - Linux/Windows/macOS| D["Install: uv pip install nemo-retriever"]
    C --> E["CUDA 13 override if needed\ntorch==2.11.0 torchvision==0.26.0\n-i https://download.pytorch.org/whl/cu130"]
    C --> F["from nemo_retriever.models.llm\nimport LiteLLMClient, LLMJudge"]
    D --> F
    F --> G[Live RAG / LLM Judge pipeline]

Loading

_{Reviews (9): Last reviewed commit: "docs(extraction): use GitHub-friendly bl..." | Re-trigger Greptile}