Make public long-document SDG docs paths portable (#269)

* Make public long-document SDG docs paths portable

* Record task162 PR status

* Fix task162 history metadata

---------

Co-authored-by: intern_nem_dev_1 <intern_nem_dev_1@intern.local>

Author: songCNMS

docs/nemotron/data/sdg/long-document.md

@@ -47,7 +47,7 @@ Each stage works in two modes:
    ```bash
    nemotron data sdg long-document ocr --batch <profile> -c 02-ocr \
        vllm_endpoint=http://compute-node:8000/v1 \
-       seed_path=/lustre/.../seed.parquet \
+       seed_path=${NEMO_RUN_DIR:-.}/output/data/sdg/long-document/seeds/seed_per_page.parquet \
        num_records=100
    ```
 
@@ -65,7 +65,7 @@ Producer stages (`ocr`, `text-qa`, `page-classification`, `visual-qa`, `single-p
 # runs the recipe against it, tears the deployment down on exit.
 nemotron data sdg long-document ocr --batch prep --serve \
     -c 02-ocr \
-    seed_path=/lustre/.../seed_per_page.parquet \
+    seed_path=${NEMO_RUN_DIR:-.}/output/data/sdg/long-document/seeds/seed_per_page.parquet \
     num_records=100
 ```
 
@@ -115,7 +115,7 @@ See the [recipe README](https://github.com/NVIDIA/nemotron/tree/main/src/nemotro
 Once the pipeline runs, the resulting parquet files can be:
 
 - Published to Hugging Face Hub as a public dataset.
-- Stored in internal Lustre and registered as a Nemotron / W&B artifact.
+- Stored in internal shared output storage and registered as a Nemotron / W&B artifact.
 - Consumed directly by training recipes via `dataset.path` or HF-dataset-id config.
 
 The recipe README has copy-pasteable templates for both publish paths.

tests/recipes/data/test_sdg_long_document_docs_portability.py

@@ -4,6 +4,7 @@
 LONG_DOCUMENT_DIR = REPO_ROOT / "src/nemotron/recipes/data/sdg/long-document"
 README = LONG_DOCUMENT_DIR / "README.md"
 OCR_CONFIG = LONG_DOCUMENT_DIR / "config/02-ocr.yaml"
+PUBLIC_DOC = REPO_ROOT / "docs/nemotron/data/sdg/long-document.md"
 PORTABLE_ROOT = "${NEMO_RUN_DIR:-.}/output/data/sdg/long-document"
 
 
@@ -12,14 +13,15 @@ def _read(path: Path) -> str:
 
 
 def test_long_document_docs_do_not_use_developer_lustre_paths() -> None:
-    for path in (README, OCR_CONFIG):
+    for path in (README, OCR_CONFIG, PUBLIC_DOC):
         text = _read(path)
         assert "/lustre/" not in text, f"{path} still contains a concrete /lustre/ example"
 
 
 def test_long_document_docs_use_portable_output_examples() -> None:
     readme_text = _read(README)
     ocr_config_text = _read(OCR_CONFIG)
+    public_doc_text = _read(PUBLIC_DOC)
 
     for suffix in (
         "seeds",
@@ -33,11 +35,13 @@ def test_long_document_docs_use_portable_output_examples() -> None:
         assert f"{PORTABLE_ROOT}/{suffix}" in readme_text
 
     assert f"{PORTABLE_ROOT}/seeds/seed_per_page.parquet" in ocr_config_text
+    assert f"{PORTABLE_ROOT}/seeds/seed_per_page.parquet" in public_doc_text
 
 
 def test_long_document_docs_preserve_cli_commands_and_options() -> None:
     readme_text = _read(README)
     ocr_config_text = _read(OCR_CONFIG)
+    public_doc_text = _read(PUBLIC_DOC)
 
     for snippet in (
         "nemotron data sdg long-document seed",
@@ -62,3 +66,12 @@ def test_long_document_docs_preserve_cli_commands_and_options() -> None:
         "num_records=100",
     ):
         assert snippet in ocr_config_text
+
+    for snippet in (
+        "nemotron data sdg long-document ocr --batch",
+        "--serve",
+        "vllm_endpoint=",
+        "seed_path=",
+        "num_records=",
+    ):
+        assert snippet in public_doc_text

workspace/interns/intern_nem_dev_1/status.md

@@ -1,12 +1,12 @@
 # intern_nem_dev_1 - 状态
 
-<!-- METADATA:STATUS=Working,TASK=task157_data_sdg_long_document_docs_path_portability_s1,ROLE=dev,SESSION=1 -->
+<!-- METADATA:STATUS=Working,TASK=task162_long_document_public_docs_path_portability_s1,ROLE=dev,SESSION=1 -->
 
 | 字段 | 值 |
 |------|-----|
 | Name | intern_nem_dev_1 |
 | Status | Working |
-| Current Task | task157_data_sdg_long_document_docs_path_portability_s1 |
-| PR | https://github.com/songCNMS/Nemotron/pull/266 |
+| Current Task | task162_long_document_public_docs_path_portability_s1 |
+| PR | https://github.com/songCNMS/Nemotron/pull/269 |
 | Session | 1 |
-| Recent Progress | Opened PR #266 for task157; long-document SDG docs/comments now use NEMO_RUN_DIR-relative examples and focused static docs checks passed |
+| Recent Progress | Opened PR #269 for task162; public long-document SDG docs now use NEMO_RUN_DIR-relative seed examples and focused static docs checks passed |

workspace/tasks/task162_long_document_public_docs_path_portability_s1/README.md

@@ -0,0 +1,21 @@
+# task162_long_document_public_docs_path_portability_s1
+
+<!-- METADATA:STATUS=InProgress,ASSIGNEE=intern_nem_dev_1,SESSION=1 -->
+
+## Scope
+
+- Replace concrete `/lustre/...` examples in `docs/nemotron/data/sdg/long-document.md` with `${NEMO_RUN_DIR:-.}/output/data/sdg/long-document/...` examples.
+- Preserve command names, option names, stage names, manual endpoint flow, `--serve` flow, and public-doc prose semantics.
+- Add focused static docs coverage for the public page.
+
+## Boundaries
+
+- No production Python, deployment YAML, CLI behavior, generated artifacts, or unrelated docs.
+- No live long-document SDG data prep, `--serve` launch, endpoint calls, downloads, train/eval, W&B, cluster jobs, deploy, artifact operations, main push, or self-merge.
+
+## Status
+
+- Base: `2cb891846c6f86d8917cd6289070c687dfdd6f91`
+- Branch: `intern_nem_dev_1/task162_long_document_public_docs_path_portability_s1`
+- PR: https://github.com/songCNMS/Nemotron/pull/269
+- Checks: focused static docs pytest, py_compile, Ruff, no-`/lustre/` grep, structured public-doc portability probe, added-line live-surface scan, and diff checks passed.

workspace/tasks/task162_long_document_public_docs_path_portability_s1/history_log.md

@@ -0,0 +1,11 @@
+# History Log
+
+<!-- METADATA:SESSION=1 -->
+
+## Session 1 - 2026-05-29
+
+- Started task162 from `origin/main` at `2cb891846c6f86d8917cd6289070c687dfdd6f91`.
+- Updated public long-document SDG docs path examples toward `${NEMO_RUN_DIR:-.}/output/data/sdg/long-document/...` portability.
+- Extended focused static docs test coverage to include the public docs page.
+- Ran focused static docs pytest, py_compile, Ruff, scoped no-`/lustre/` grep, structured public-doc probe, added-line live-surface scan, and `git diff --check`.
+- Opened PR #269: https://github.com/songCNMS/Nemotron/pull/269

workspace/tasks/task162_long_document_public_docs_path_portability_s1/task_knowledge.md

@@ -0,0 +1,6 @@
+# Task Knowledge
+
+<!-- METADATA:STATUS=Working,TASK=task162_long_document_public_docs_path_portability_s1,ROLE=dev,SESSION=1 -->
+
+- Public long-document SDG docs should use `${NEMO_RUN_DIR:-.}/output/data/sdg/long-document/...` examples and avoid concrete cluster/user storage paths.
+- The task is docs/static-test only; manual endpoint and `--serve` examples must remain recognizable with the same CLI options.