codemaivanngu commited on
Commit
fbe2a3e
·
verified ·
1 Parent(s): 925e2c5

Publish 3e536ca project-local runtime and inherited training config

Browse files
INSTALL.md CHANGED
@@ -212,6 +212,12 @@ To reuse the current training environment, invoke the versioned
212
  data, and delegates cache reuse to the existing `copy_gym_venvs.sh`, then checks
213
  imports. It preserves the launcher's environment layout and starts no training.
214
 
 
 
 
 
 
 
215
  For the pinned public dataset revision
216
  `6f274cd4cbc27e656d7ddae7c3b591669866dc84`,
217
  [`packaging/prepare_dataset.py`](packaging/prepare_dataset.py) combines extraction,
 
212
  data, and delegates cache reuse to the existing `copy_gym_venvs.sh`, then checks
213
  imports. It preserves the launcher's environment layout and starts no training.
214
 
215
+ When an execution scope excludes the existing system cache, use
216
+ `packaging/prepare_project_runtime.sh` to copy the required cached environments
217
+ into a fresh project directory. It generates an inherited trainer YAML that
218
+ changes only `env.nemo_gym.uv_venv_dir`, and verifies existing source/data and
219
+ frontend imports without editing the base trainer configuration or system cache.
220
+
221
  For the pinned public dataset revision
222
  `6f274cd4cbc27e656d7ddae7c3b591669866dc84`,
223
  [`packaging/prepare_dataset.py`](packaging/prepare_dataset.py) combines extraction,
SOURCE_RELEASE.json CHANGED
@@ -1,11 +1,11 @@
1
  {
2
- "archive": "spreadsheet-gym-integration-9811a5177507.tar.gz",
3
- "sha256": "c355a045c1faa60ae5bdc76a5960b6dec167292b682334908e7e87e7b1e82279",
4
- "bytes": 90610,
5
- "source_commit": "9811a5177507b09b14590c3c58477cd79fc655f6",
6
- "files": 51,
7
  "branch": "tungks/feat/tool/spreadsheet-rl",
8
  "reproducible_rebuild": true,
9
  "archive_manifest_verified": true,
10
- "validation": "83 passed, 4 Linux/XLSX tests skipped on Mac; existing controller reached from train with HTTP 200; frontend cache preparation pending"
11
  }
 
1
  {
2
+ "archive": "spreadsheet-gym-integration-3e536cae8dfa.tar.gz",
3
+ "sha256": "d1a37c134043ae03eba68a6bb4a7bb0f7e6aecadfe563ce90338e5b9f5298bed",
4
+ "bytes": 94271,
5
+ "source_commit": "3e536cae8dfa74805da4dfd33a9874e570e7fd80",
6
+ "files": 54,
7
  "branch": "tungks/feat/tool/spreadsheet-rl",
8
  "reproducible_rebuild": true,
9
  "archive_manifest_verified": true,
10
+ "validation": "86 passed, 4 Linux/XLSX tests skipped on Mac; project-local runtime deployment pending"
11
  }
SOURCE_TRANSPORT.json CHANGED
@@ -1,11 +1,11 @@
1
  {
2
  "schema_version": 1,
3
- "source_commit": "9811a5177507b09b14590c3c58477cd79fc655f6",
4
  "encoding": "base64",
5
- "encoded_file": "spreadsheet-source-9811a5177507.base64.txt",
6
- "encoded_bytes": 120817,
7
- "encoded_sha256": "07bee84e4325d049a2614c2e86cf4d88837081475d54c1f483cceefe24df38dd",
8
- "archive": "spreadsheet-gym-integration-9811a5177507.tar.gz",
9
- "archive_bytes": 90610,
10
- "archive_sha256": "c355a045c1faa60ae5bdc76a5960b6dec167292b682334908e7e87e7b1e82279"
11
  }
 
1
  {
2
  "schema_version": 1,
3
+ "source_commit": "3e536cae8dfa74805da4dfd33a9874e570e7fd80",
4
  "encoding": "base64",
5
+ "encoded_file": "spreadsheet-source-3e536cae8dfa.base64.txt",
6
+ "encoded_bytes": 125697,
7
+ "encoded_sha256": "20021049ea4850a3ff7de261423bee4598118d8d2587c11ebe5b2492f4e9f808",
8
+ "archive": "spreadsheet-gym-integration-3e536cae8dfa.tar.gz",
9
+ "archive_bytes": 94271,
10
+ "archive_sha256": "d1a37c134043ae03eba68a6bb4a7bb0f7e6aecadfe563ce90338e5b9f5298bed"
11
  }
TRAINING_SETUP.md CHANGED
@@ -15,6 +15,7 @@ paths and private keys stay outside public release artifacts.
15
  | `start_service.py` | Existing runtime bundle, qualified catalog, node-local service root and sandbox root receipt; optional `--export-private-config` | Start/reuse the API; export its Gym override to shared storage with mode 0600 |
16
  | `prepare_training_node.sh` | `TRAINER_ROOT NLTK_DATA TRAINER_CONFIG` followed by the Python options below | Use the existing activation/cache helper, install source/config, convert data and check existing frontend venv imports |
17
  | `prepare_cached_frontends.py` | `--venv-root`, `--expected-hostname`, `--receipt` | Copy the existing `math_with_judge` and `simple_agent` caches for the two new frontends; no installs or venv aliases |
 
18
  | `build_source_bundle.py` | Clean source checkout; `--output-dir` outside it | Small source archive and checksum, pinned to the exact Git commit |
19
 
20
  Conversion, source installation and source packaging use only Python's standard
@@ -94,9 +95,32 @@ copies the existing `math_with_judge` cache for `spreadsheet`, and the existing
94
  Complete existing destinations are reused; linked or incomplete destinations
95
  are preserved and rejected. This creates real cache copies without installing
96
  dependencies or modifying the project's helper. On a Bridge binding whose write
97
- root excludes the cache directory, an operator must run this standalone step
98
- first. It writes an owner-only receipt and must not be used to bypass the
99
- binding's filesystem policy.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
100
 
101
  Each invocation saves a private phase receipt and cache-helper log under a
102
  unique setup directory. `TRAINING_NODE_PREPARED` means these setup checks passed;
 
15
  | `start_service.py` | Existing runtime bundle, qualified catalog, node-local service root and sandbox root receipt; optional `--export-private-config` | Start/reuse the API; export its Gym override to shared storage with mode 0600 |
16
  | `prepare_training_node.sh` | `TRAINER_ROOT NLTK_DATA TRAINER_CONFIG` followed by the Python options below | Use the existing activation/cache helper, install source/config, convert data and check existing frontend venv imports |
17
  | `prepare_cached_frontends.py` | `--venv-root`, `--expected-hostname`, `--receipt` | Copy the existing `math_with_judge` and `simple_agent` caches for the two new frontends; no installs or venv aliases |
18
+ | `prepare_project_runtime.sh` | `TRAINER_ROOT NLTK_DATA`, then `--base-config`, `--runtime-root`, `--qualified-dir`, `--output-dir`, `--private-config`, `--expected-hostname` | Prepare project-local caches and an inherited YAML; verify existing integration/data without modifying their files |
19
  | `build_source_bundle.py` | Clean source checkout; `--output-dir` outside it | Small source archive and checksum, pinned to the exact Git commit |
20
 
21
  Conversion, source installation and source packaging use only Python's standard
 
95
  Complete existing destinations are reused; linked or incomplete destinations
96
  are preserved and rejected. This creates real cache copies without installing
97
  dependencies or modifying the project's helper. On a Bridge binding whose write
98
+ root excludes the cache directory, use the project-local runtime below instead
99
+ of writing outside the binding policy. The standalone cache copier also accepts
100
+ `--source-root`, `--include-model`, and `--relocate-activation` for copying an
101
+ existing cache into a new location.
102
+
103
+ ## Project-local runtime under an existing Bridge scope
104
+
105
+ `prepare_project_runtime.sh` sources the existing import activation and invokes
106
+ `prepare_project_runtime.py`. The active trainer must already have the matching
107
+ spreadsheet frontend sources, config paths, and verified converted data. It
108
+ checks these prerequisites without installing source, patching the base YAML,
109
+ recalculating workbooks, or recreating data.
110
+
111
+ All persistent writes are below the fresh `--runtime-root`. It copies existing
112
+ `math_with_judge`, `simple_agent`, and `vllm_model` caches from
113
+ `--source-cache-root` (default `/opt/gym_venvs`) into the project's resource,
114
+ agent, and model cache directories. Only copied Bash activation scripts are
115
+ retargeted. Import probes require the copied interpreter prefixes and matching
116
+ Ray/OpenAI versions. Python bytecode writes are disabled during preparation and
117
+ probes.
118
+
119
+ The new `spreadsheet.yaml` inherits `--base-config` and overrides only
120
+ `env.nemo_gym.uv_venv_dir`. Setup verifies the complete effective configuration
121
+ equals the base with that single change. Use this generated YAML when launching
122
+ training. `setup.json` pins the base/config/source/data hashes and live service
123
+ health. `TRAINING_NODE_PREPARED` confirms preparation; no training is launched.
124
 
125
  Each invocation saves a private phase receipt and cache-helper log under a
126
  unique setup directory. `TRAINING_NODE_PREPARED` means these setup checks passed;
spreadsheet-gym-integration-3e536cae8dfa.tar.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d1a37c134043ae03eba68a6bb4a7bb0f7e6aecadfe563ce90338e5b9f5298bed
3
+ size 94271
spreadsheet-gym-integration-3e536cae8dfa.tar.gz.sha256 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31b1cf469f993525a593d5e22d6c891fcede70f631685ae2a4a8c49d5ea29788
3
+ size 114
spreadsheet-source-3e536cae8dfa.base64.txt ADDED
The diff for this file is too large to render. See raw diff