nshtrainer 0.11.10__py3-none-any.whl → 0.11.12__py3-none-any.whl

nshtrainer/_checkpoint/metadata.py

@@ -137,7 +137,11 @@ def _link_checkpoint_metadata(checkpoint_path: Path, linked_checkpoint_path: Pat
     linked_path = linked_checkpoint_path.with_suffix(METADATA_PATH_SUFFIX)
     try:
         try:
-            linked_path.symlink_to(path)
+            # linked_path.symlink_to(path)
+            # We should store the path as a relative path
+            # to the metadata file to avoid issues with
+            # moving the checkpoint directory
+            linked_path.symlink_to(path.relative_to(linked_path.parent))
         except OSError:
             # on Windows, special permissions are required to create symbolic links as a regular user
             # fall back to copying the file

nshtrainer/_checkpoint/saver.py

@@ -12,22 +12,25 @@ def _link_checkpoint(
     linkpath: str | Path | os.PathLike,
     *,
     metadata: bool,
+    remove_existing: bool = True,
 ):
     if not isinstance(filepath, Path):
         filepath = Path(filepath)
     if not isinstance(linkpath, Path):
         linkpath = Path(linkpath)
 
-    if linkpath.exists():
-        if linkpath.is_symlink() or linkpath.is_file():
-            linkpath.unlink()
-        elif linkpath.is_dir():
-            shutil.rmtree(linkpath)
-        _remove_checkpoint_metadata(linkpath)
+    if remove_existing:
+        if linkpath.exists():
+            if linkpath.is_symlink() or linkpath.is_file():
+                linkpath.unlink()
+            elif linkpath.is_dir():
+                shutil.rmtree(linkpath)
+
+        if metadata:
+            _remove_checkpoint_metadata(linkpath)
 
     try:
-        target_path = filepath.relative_to(linkpath.parent)
-        linkpath.symlink_to(target_path)
+        linkpath.symlink_to(filepath.relative_to(linkpath.parent))
     except OSError:
         # on Windows, special permissions are required to create symbolic links as a regular user
         # fall back to copying the file

nshtrainer/callbacks/checkpoint/_base.py

@@ -62,6 +62,7 @@ class CheckpointBase(Checkpoint, ABC, Generic[TConfig]):
 
         self.config = config
         self.dirpath = dirpath / self.name()
+        self.dirpath.mkdir(parents=True, exist_ok=True)
         self.symlink_dirpath = dirpath
 
         self._last_global_step_saved = 0

nshtrainer/callbacks/checkpoint/on_exception_checkpoint.py

@@ -73,6 +73,9 @@ class OnExceptionCheckpoint(_OnExceptionCheckpoint):
         self.config = config
         del config
 
+        dirpath = dirpath / "on_exception"
+        dirpath.mkdir(parents=True, exist_ok=True)
+
         super().__init__(dirpath, filename)
 
     @property

nshtrainer/model/config.py

@@ -3,7 +3,6 @@ import logging
 import os
 import string
 import time
-import warnings
 from abc import ABC, abstractmethod
 from collections.abc import Iterable, Sequence
 from datetime import timedelta
@@ -50,10 +49,6 @@ from ..util._environment_info import EnvironmentConfig
 log = logging.getLogger(__name__)
 
 
-class IdSeedWarning(Warning):
-    pass
-
-
 class BaseProfilerConfig(C.Config, ABC):
     dirpath: str | Path | None = None
     """
@@ -1478,35 +1473,12 @@ class BaseConfig(C.Config):
     _rng: ClassVar[np.random.Generator | None] = None
 
     @staticmethod
-    def generate_id(
-        *,
-        length: int = 8,
-        ignore_rng: bool = False,
-    ) -> str:
+    def generate_id(*, length: int = 8) -> str:
         """
         Generate a random ID of specified length.
 
-        Args:
-            length (int): The length of the generated ID. Default is 8.
-            ignore_rng (bool): If True, ignore the global random number generator and use a new one. Default is False.
-
-        Returns:
-            str: The generated random ID.
-
-        Raises:
-            IdSeedWarning: If the global random number generator is None and ignore_rng is False.
-
-        Notes:
-            - The generated IDs will not be reproducible if the global random number generator is None and ignore_rng is False.
-            - To ensure reproducibility, call BaseConfig.set_seed(...) before generating any IDs.
         """
-        rng = BaseConfig._rng if not ignore_rng else np.random.default_rng()
-        if rng is None:
-            warnings.warn(
-                "BaseConfig._rng is None. The generated IDs will not be reproducible. "
-                + "To fix this, call BaseConfig.set_seed(...) before generating any IDs.",
-                category=IdSeedWarning,
-            )
+        if (rng := BaseConfig._rng) is None:
             rng = np.random.default_rng()
 
         alphabet = list(string.ascii_lowercase + string.digits)

{nshtrainer-0.11.10.dist-info → nshtrainer-0.11.12.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: nshtrainer
-Version: 0.11.10
+Version: 0.11.12
 Summary: 
 Author: Nima Shoghi
 Author-email: nimashoghi@gmail.com

{nshtrainer-0.11.10.dist-info → nshtrainer-0.11.12.dist-info}/RECORD

@@ -1,7 +1,7 @@
 nshtrainer/__init__.py,sha256=39loiLLXbaGiozEsAn8mPHopxaPsek8JsgR9DD2gxtY,583
 nshtrainer/_checkpoint/loader.py,sha256=DSaNR8194kWon4O1svslNsCcN_8vlyLbF0LNCPfUpzI,13789
-nshtrainer/_checkpoint/metadata.py,sha256=n2PMGdA3Jn3BuoyDXVCF9dUnNjuuru5CL9jqMD-X4Vk,4918
-nshtrainer/_checkpoint/saver.py,sha256=TuSAP39DOOVvSnSukQ9RitMV60JnDg6L27fMRc2uVJc,1358
+nshtrainer/_checkpoint/metadata.py,sha256=onmetLp5eKbA86abq1PTkwAOO7bWj7Pa1EGUjl2TEjQ,5153
+nshtrainer/_checkpoint/saver.py,sha256=DkbCH0YeOJ71m32vAARiQdGBf0hvwwdoAV8LOFGy-0Y,1428
 nshtrainer/_experimental/__init__.py,sha256=2tQIcrWT8U8no_AeBTYnozaTmxN40kuAJdGQ4b-PoWM,120
 nshtrainer/_experimental/flops/__init__.py,sha256=edo9Ez3LlrnxkNRX9W6YBhPkRPKYGLpkpnl5gx7sEX8,1550
 nshtrainer/_experimental/flops/flop_counter.py,sha256=-sL0Fy6poXa__hyzUMdZScjPULp4coQELQpPU6p6dXU,25736
@@ -11,10 +11,10 @@ nshtrainer/callbacks/_throughput_monitor_callback.py,sha256=aJo_11rc4lo0IYOd-kHm
 nshtrainer/callbacks/actsave.py,sha256=qbnaKts4_dvjPeAaPtv7Ds12_vEWzaHUfg_--49NB9I,4041
 nshtrainer/callbacks/base.py,sha256=UnlYZAqSb8UwBJR-N5-XunxFx2yZjZ4lyGqUfhbCRlI,3555
 nshtrainer/callbacks/checkpoint/__init__.py,sha256=g-3zIthupERKqWZQw-A_busQPaPRkto6iHBV-M7nK1Y,527
-nshtrainer/callbacks/checkpoint/_base.py,sha256=9HQSa-toOyjtuDldQ71gaDVBRdryAaB_nRv5Y554tIk,5938
+nshtrainer/callbacks/checkpoint/_base.py,sha256=wb5ARqqSslZXB5FSwNd9a_7qtt4-F9dC7E1vNHKCK1o,5994
 nshtrainer/callbacks/checkpoint/best_checkpoint.py,sha256=ySeyALxc-YJaS5IhmW0MkAhr41Mxsm_BllHuufXGy4Y,2067
 nshtrainer/callbacks/checkpoint/last_checkpoint.py,sha256=CM8f37dwaYHkjQFfJNTZTzSoF45zEjFRm-Fg1CzYmP4,1037
-nshtrainer/callbacks/checkpoint/on_exception_checkpoint.py,sha256=s8tOHrnb_uVqLVeV2K38ZszXrXPTEGdDVfXuXgo_KDQ,3277
+nshtrainer/callbacks/checkpoint/on_exception_checkpoint.py,sha256=ctT88EGT22_t_6tr5r7Sfo43cuve6XeroBnBYRMPOus,3372
 nshtrainer/callbacks/early_stopping.py,sha256=LGn3rdbvkFfUo9kwMzK4eMGlPAqD9uFdowDx6VdfozQ,3761
 nshtrainer/callbacks/ema.py,sha256=8-WHmKFP3VfnzMviJaIFmVD9xHPqIPmq9NRF5xdu3c8,12131
 nshtrainer/callbacks/finite_checks.py,sha256=gJC_RUr3ais3FJI0uB6wUZnDdE3WRwCix3ppA3PwQXA,2077
@@ -54,7 +54,7 @@ nshtrainer/metrics/__init__.py,sha256=ObLIELGguIEcUpRsUkqh1ltrvZii6vglTpJGrPvoy0
 nshtrainer/metrics/_config.py,sha256=jgRBfDAQLFTW7AiUY7CRtdfts6CR6keeuqm0FFMWCzQ,1288
 nshtrainer/model/__init__.py,sha256=RlGW5a46DZcqK6cYICYxDaKpZIEj-8zLxoMrl432tno,1429
 nshtrainer/model/base.py,sha256=AXRfEsFAT0Ln7zjYVPU5NgtHS_c8FZM-M4pyLamO7OA,17516
-nshtrainer/model/config.py,sha256=F-doUiqPVw4yepT6FRqhflEiiMEWr_PuFC6lKzFWktA,53809
+nshtrainer/model/config.py,sha256=3qSaXNjpKnxF60LbfpONsVhjtrteLEdHkjb5KAhAnIk,52757
 nshtrainer/model/modules/callback.py,sha256=K0-cyEtBcQhI7Q2e-AGTE8T-GghUPY9DYmneU6ULV6g,6401
 nshtrainer/model/modules/debug.py,sha256=Yy7XEdPou9BkCsD5hJchwJGmCVGrfUru5g9VjPM4uAw,1120
 nshtrainer/model/modules/distributed.py,sha256=ABpR9d-3uBS_fivfy_WYW-dExW6vp5BPaoPQnOudHng,1725
@@ -82,6 +82,6 @@ nshtrainer/util/seed.py,sha256=Or2wMPsnQxfnZ2xfBiyMcHFIUt3tGTNeMMyOEanCkqs,280
 nshtrainer/util/slurm.py,sha256=rofIU26z3SdL79SF45tNez6juou1cyDLz07oXEZb9Hg,1566
 nshtrainer/util/typed.py,sha256=NGuDkDzFlc1fAoaXjOFZVbmj0mRFjsQi1E_hPa7Bn5U,128
 nshtrainer/util/typing_utils.py,sha256=8ptjSSLZxlmy4FY6lzzkoGoF5fGNClo8-B_c0XHQaNU,385
-nshtrainer-0.11.10.dist-info/METADATA,sha256=KC2AcgRB55-p9SZvLv_MD7wATBOSQJp4VcHzU8LfiCI,861
-nshtrainer-0.11.10.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-nshtrainer-0.11.10.dist-info/RECORD,,
+nshtrainer-0.11.12.dist-info/METADATA,sha256=Lthmwj2EpbkCNWlDg2n7z4zS197Pn5Kt1aUMayB70oo,861
+nshtrainer-0.11.12.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+nshtrainer-0.11.12.dist-info/RECORD,,