Skip to content

Commit f39a05f

Browse files
songCNMSintern_nem_dev_1
andauthored
[task154][intern_nem_dev_1] Make Super3 RL data-prep doc output portable (#261)
* Make Super3 RL data-prep doc output portable * Record task154 PR status --------- Co-authored-by: intern_nem_dev_1 <intern_nem_dev_1@intern.local>
1 parent 795eb92 commit f39a05f

6 files changed

Lines changed: 71 additions & 7 deletions

File tree

docs/nemotron/super3/rl/data-prep.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -100,7 +100,7 @@ with `responses_create_params.input`, `question`, `expected_answer`,
100100
```bash
101101
source /work-agents/.venv/bin/activate
102102
python src/nemotron/recipes/super3/milestones/m0_data_env/prepare_m0_assets.py \
103-
--output-dir /mnt/3fs/data/lei.song/nemotron/m0_data_env_foundation/smoke-20260516 \
103+
--output-dir "${NEMO_RUN_DIR:-.}/output/super3/m0_data_env_foundation/smoke-20260516" \
104104
--max-train-per-dataset 100 \
105105
--max-val-per-dataset 25
106106
```

tests/recipes/super3/test_m0_data_env.py

Lines changed: 13 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -40,8 +40,12 @@
4040
)
4141
RL_RAW_BLEND_PATH = REPO_ROOT / "src/nemotron/recipes/super3/stage2_rl/config/data_prep/data_blend_raw.json"
4242
M0_DATA_ENV_README_PATH = REPO_ROOT / "src/nemotron/recipes/super3/milestones/m0_data_env/README.md"
43+
SUPER3_RL_DATA_PREP_DOC_PATH = REPO_ROOT / "docs/nemotron/super3/rl/data-prep.md"
4344
OLD_SUPER3_RL_BLEND_SLUG = "nvidia/Nemotron-3-Super-RL-Training-Blends"
4445
LIVE_SUPER3_RL_BLEND_SLUG = "nvidia/Nemotron-RL-Super-Training-Blends"
46+
PORTABLE_M0_SMOKE_OUTPUT_DIR = (
47+
'"${NEMO_RUN_DIR:-.}/output/super3/m0_data_env_foundation/smoke-20260516"'
48+
)
4549

4650

4751
def _require_live_hf_tests() -> None:
@@ -103,8 +107,15 @@ def test_m0_data_env_readme_uses_portable_output_examples() -> None:
103107
readme = M0_DATA_ENV_README_PATH.read_text()
104108

105109
assert "/mnt/3fs/data/lei.song/nemotron" not in readme
106-
assert '--output-dir "${NEMO_RUN_DIR:-.}/output/super3/m0_data_env_foundation/smoke-20260516"' in readme
107-
assert '--input-dir "${NEMO_RUN_DIR:-.}/output/super3/m0_data_env_foundation/smoke-20260516"' in readme
110+
assert f"--output-dir {PORTABLE_M0_SMOKE_OUTPUT_DIR}" in readme
111+
assert f"--input-dir {PORTABLE_M0_SMOKE_OUTPUT_DIR}" in readme
112+
113+
114+
def test_super3_rl_data_prep_doc_uses_portable_m0_output_example() -> None:
115+
doc = SUPER3_RL_DATA_PREP_DOC_PATH.read_text()
116+
117+
assert "/mnt/3fs/data/lei.song/nemotron" not in doc
118+
assert f"--output-dir {PORTABLE_M0_SMOKE_OUTPUT_DIR}" in doc
108119

109120

110121
def test_super3_rl_blend_slug_resolves_on_hf_when_network_available() -> None:
Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -1,12 +1,12 @@
11
# intern_nem_dev_1 - 状态
22

3-
<!-- METADATA:STATUS=Working,TASK=task150_super3_stage1_sft_tiny_blend_contract_s1,ROLE=independent,SESSION=1 -->
3+
<!-- METADATA:STATUS=Working,TASK=task154_super3_rl_data_prep_doc_output_portability_s1,ROLE=independent,SESSION=1 -->
44

55
| 字段 ||
66
|------|-----|
77
| Name | intern_nem_dev_1 |
88
| Status | Working |
9-
| Current Task | task150_super3_stage1_sft_tiny_blend_contract_s1 |
10-
| PR | https://github.com/songCNMS/Nemotron/pull/258 |
9+
| Current Task | task154_super3_rl_data_prep_doc_output_portability_s1 |
10+
| PR | https://github.com/songCNMS/Nemotron/pull/261 |
1111
| Session | 1 |
12-
| Recent Progress | Opened PR #258 for task150; Super3 Stage1 SFT tiny uses Super3-owned non-empty blend and null used_in_filter; focused checks passed |
12+
| Recent Progress | Opened PR #261 for task154; Super3 RL data-prep doc M0 output example now uses NEMO_RUN_DIR-relative path and focused docs checks passed |
Lines changed: 24 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,24 @@
1+
# task154_super3_rl_data_prep_doc_output_portability_s1
2+
3+
<!-- METADATA:STATUS=InProgress,ASSIGNEE=intern_nem_dev_1 -->
4+
5+
## Scope
6+
7+
- Update `docs/nemotron/super3/rl/data-prep.md` so the M0 data-env
8+
`--output-dir` example uses a portable `${NEMO_RUN_DIR:-.}`-relative
9+
Super3 output path instead of a developer-local `/mnt/3fs/...` path.
10+
- Add focused static test coverage proving the stale named-user path is absent
11+
and the portable example is present.
12+
13+
## Boundaries
14+
15+
- Static docs/test/docs-only.
16+
- No live M0 data prep, HF/dataset download, health baseline, train/eval,
17+
endpoint, W&B, cluster, deploy, artifact download, direct `main`/`master`
18+
push, or self-merge.
19+
20+
## Status
21+
22+
- Branch: `intern_nem_dev_1/task154_super3_rl_data_prep_doc_output_portability_s1`
23+
- Base: `795eb92359257ed82816a8685db0f9cae1c751ae`
24+
- PR: https://github.com/songCNMS/Nemotron/pull/261
Lines changed: 18 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,18 @@
1+
# task154_super3_rl_data_prep_doc_output_portability_s1 history
2+
3+
<!-- METADATA:SESSION=1 -->
4+
5+
## Session 1 - 2026-05-29
6+
7+
- Accepted PM assignment for task154 from peer message.
8+
- Created branch
9+
`intern_nem_dev_1/task154_super3_rl_data_prep_doc_output_portability_s1`
10+
from `origin/main` at `795eb92359257ed82816a8685db0f9cae1c751ae`.
11+
- Updated `docs/nemotron/super3/rl/data-prep.md` to use
12+
`"${NEMO_RUN_DIR:-.}/output/super3/m0_data_env_foundation/smoke-20260516"`
13+
for the M0 `--output-dir` example.
14+
- Extended focused M0 docs static coverage in
15+
`tests/recipes/super3/test_m0_data_env.py`.
16+
- Ran focused docs pytest, py_compile, Ruff, scoped product-doc grep,
17+
added-line live-surface scan, and diff checks.
18+
- Opened PR #261 to `main`: https://github.com/songCNMS/Nemotron/pull/261.
Lines changed: 11 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,11 @@
1+
# task154_super3_rl_data_prep_doc_output_portability_s1 knowledge
2+
3+
<!-- METADATA:SESSION=1 -->
4+
5+
## Working Notes
6+
7+
- Super3 RL product docs should not include named-user `/mnt/3fs` output
8+
examples.
9+
- M0 data-env examples should use
10+
`"${NEMO_RUN_DIR:-.}/output/super3/m0_data_env_foundation/smoke-20260516"`
11+
so they work outside a specific developer workspace.

0 commit comments

Comments
 (0)