fix: update customizer docs based on QA feedback

[mckornfield]

Summary by CodeRabbit

• Documentation
  • Updated embedding customization guides and notebook/tutorial assets to consistently use nvidia/llama-nemotron-embed-1b-v2.
  • Clarified embedding training options: set training.finetuning_type to "lora_merged"; unmerged LoRA adapters aren’t supported (ONNX requirement).
  • Reworked the private HuggingFace import tutorial to use the NeMo Platform SDK workflow, with updated prerequisites and inference validation.
  • Updated SFT evaluation examples to call v1/completions using prompt, and adjusted output parsing accordingly.
  • Refreshed configuration/model-learning diagrams and LoRA adapter examples; fixed a minor tutorial typo.
  • Removed %%bash directives from relevant notebook dependency-install cells.

[github-actions]

🌿 Preview your docs: https://nvidia-preview-cust-docs-fixes-mck.docs.buildwithfern.com/nemo-platform

[coderabbitai]

Note

Reviews paused

It looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the reviews.auto_review.auto_pause_after_reviewed_commits setting.

Use the following commands to manage reviews:

• @coderabbitai resume to resume automatic reviews.
• @coderabbitai review to trigger a single review.

Use the checkboxes below for quick actions:

• ▶️ Resume reviews
• 🔍 Trigger review

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: a46a41cd-4d83-4f32-807c-d929baee9c96

📥 Commits

Reviewing files that changed from the base of the PR and between 1ab9baa and cfd68a2.

📒 Files selected for processing (1)

• docs/customizer/tutorials/import-hf-model.mdx

💤 Files with no reviewable changes (1)

• docs/customizer/tutorials/import-hf-model.mdx

---

📝 Walkthrough

Walkthrough

Docs and notebook examples were updated for embedding naming, SDK-based import and fine-tuning flows, completions-based evaluation calls, adapter API names, and throughput install snippets. Tutorial index copy was also corrected.

Changes

Customizer documentation updates

Layer / File(s)	Summary
Embedding guidance and tutorial naming docs/customizer/models/embedding.mdx, docs/customizer/tutorials/embedding-customization-job.*, docs/fern/components/notebooks/embedding-customization-job.*	nvidia/llama-nemotron-embed-1b-v2 replaces the older embedding model name in the tutorial and generated notebook docs, and the model note adds the training.finetuning_type: "lora_merged" requirement.
Import-HF prerequisites and SDK bootstrap docs/customizer/tutorials/import-hf-model.mdx	The import tutorial prerequisites now center on the NeMo Platform SDK, and the setup section adds SDK initialization, secret storage, and model/fileset creation flow.
Import-HF training, deployment, and test docs/customizer/tutorials/import-hf-model.mdx	The tutorial submits and polls the Automodel LoRA job, creates and waits on the LoRA-enabled deployment, and updates the inference examples and next-step link.
SFT evaluation endpoint docs/customizer/tutorials/sft-customization-job.*, docs/fern/components/notebooks/sft-customization-job.*	The SFT evaluation examples and generated notebook docs now use v1/completions with a single prompt and read output from choices[0].text; the setup sentence adds a period.
Configuration tutorial API and diagram syntax docs/customizer/tutorials/understand-configurations-and-models.mdx	The tutorial intro, Mermaid labels, and adapter examples now use client.models.adapters.* with updated argument names.
Throughput install snippet cleanup docs/customizer/tutorials/optimize-throughput.*, docs/fern/components/notebooks/optimize-throughput.*	The dependency-install snippets remove %%bash while keeping the same conditional uv-versus-pip shell logic.
Tutorial index copy docs/customizer/tutorials/index.mdx	The tutorial index entry text is corrected to say modifying all training parameters.

Suggested labels

docs

Suggested reviewers

• soluwalana
• gabwow

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately summarizes the main change: QA-driven updates to the customizer docs.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch cust-docs-fixes/mck

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 3

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@docs/customizer/tutorials/import-hf-model.mdx`:
- Around line 167-169: The change to base_model.spec.linear_layers is only
applied to the local response object and will not affect the LoRA job that uses
default/{base_model.name}; update the actual model record instead. In the
tutorial flow around base_model.spec and client.models.update(...), persist the
spec change through the client before proceeding, or remove this local mutation
if it is not meant to be saved.

In `@docs/customizer/tutorials/understand-configurations-and-models.mdx`:
- Around line 189-195: The tutorial is still using the old nested adapter
methods instead of the published adapter APIs. Update the examples in this
document to use client.models.update_adapter(adapter_name=...) in place of
client.models.adapters.update, and client.models.create_adapter(name=...) in
place of client.models.adapters.create, keeping the surrounding parameters and
behavior the same. Use the client.models.adapters and client.models symbols to
find every affected snippet in this tutorial and replace them consistently.

In `@docs/fern/components/notebooks/optimize-throughput.json`:
- Around line 36-38: The notebook export is hardcoding all code cells as Python,
which causes shell cells like the one in optimize-throughput to be mislabeled.
Update `docs/fern/scripts/ipynb-to-fern-json.py` so it preserves each cell’s
source language when generating the `language` field, using the notebook cell
metadata for the shellscript cell and keeping Python only for actual Python
cells. Make sure the export logic that builds the JSON for notebook cells
carries this language through into
`docs/fern/components/notebooks/optimize-throughput.json`.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: ad42ea1c-58d6-4e87-ba77-9b73f98a4798

📥 Commits

Reviewing files that changed from the base of the PR and between 4416dc5 and b40c329.

📒 Files selected for processing (16)

• docs/customizer/models/embedding.mdx
• docs/customizer/tutorials/embedding-customization-job.ipynb
• docs/customizer/tutorials/embedding-customization-job.mdx
• docs/customizer/tutorials/import-hf-model.mdx
• docs/customizer/tutorials/index.mdx
• docs/customizer/tutorials/optimize-throughput.ipynb
• docs/customizer/tutorials/optimize-throughput.mdx
• docs/customizer/tutorials/sft-customization-job.ipynb
• docs/customizer/tutorials/sft-customization-job.mdx
• docs/customizer/tutorials/understand-configurations-and-models.mdx
• docs/fern/components/notebooks/embedding-customization-job.json
• docs/fern/components/notebooks/embedding-customization-job.ts
• docs/fern/components/notebooks/optimize-throughput.json
• docs/fern/components/notebooks/optimize-throughput.ts
• docs/fern/components/notebooks/sft-customization-job.json
• docs/fern/components/notebooks/sft-customization-job.ts

💤 Files with no reviewable changes (2)

• docs/customizer/tutorials/optimize-throughput.mdx
• docs/customizer/tutorials/optimize-throughput.ipynb

[github-actions]

Suite	Lines Covered	Line Rate	Branch Rate
Unit Tests	21322/27924	76.4%	61.4%
Integration Tests	12353/26693	46.3%	19.7%

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@docs/customizer/tutorials/import-hf-model.mdx`:
- Around line 149-151: The ConflictError branch in the HuggingFace import flow
currently only retrieves the existing fileset, which leaves its stored config
stale across reruns. Update the logic in the exception handler around MODEL_NAME
to modify the existing fileset instead of just calling
client.files.filesets.retrieve, so any changed HuggingFace settings like repo_id
or token_secret are written back consistently.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 7a051992-b470-4301-b57d-66aa3fc30774

📥 Commits

Reviewing files that changed from the base of the PR and between b40c329 and ec38fe9.

📒 Files selected for processing (16)

• docs/customizer/models/embedding.mdx
• docs/customizer/tutorials/embedding-customization-job.ipynb
• docs/customizer/tutorials/embedding-customization-job.mdx
• docs/customizer/tutorials/import-hf-model.mdx
• docs/customizer/tutorials/index.mdx
• docs/customizer/tutorials/optimize-throughput.ipynb
• docs/customizer/tutorials/optimize-throughput.mdx
• docs/customizer/tutorials/sft-customization-job.ipynb
• docs/customizer/tutorials/sft-customization-job.mdx
• docs/customizer/tutorials/understand-configurations-and-models.mdx
• docs/fern/components/notebooks/embedding-customization-job.json
• docs/fern/components/notebooks/embedding-customization-job.ts
• docs/fern/components/notebooks/optimize-throughput.json
• docs/fern/components/notebooks/optimize-throughput.ts
• docs/fern/components/notebooks/sft-customization-job.json
• docs/fern/components/notebooks/sft-customization-job.ts

💤 Files with no reviewable changes (2)

• docs/customizer/tutorials/optimize-throughput.mdx
• docs/customizer/tutorials/optimize-throughput.ipynb

✅ Files skipped from review due to trivial changes (5)

• docs/customizer/tutorials/index.mdx
• docs/customizer/models/embedding.mdx
• docs/customizer/tutorials/embedding-customization-job.ipynb
• docs/customizer/tutorials/understand-configurations-and-models.mdx
• docs/customizer/tutorials/embedding-customization-job.mdx

🚧 Files skipped from review as they are similar to previous changes (8)

• docs/customizer/tutorials/sft-customization-job.ipynb
• docs/fern/components/notebooks/sft-customization-job.ts
• docs/customizer/tutorials/sft-customization-job.mdx
• docs/fern/components/notebooks/embedding-customization-job.ts
• docs/fern/components/notebooks/optimize-throughput.json
• docs/fern/components/notebooks/embedding-customization-job.json
• docs/fern/components/notebooks/optimize-throughput.ts
• docs/fern/components/notebooks/sft-customization-job.json