wandb 0.19.1rc1__py3-none-win32.whl → 0.19.2__py3-none-win32.whl

wandb/__init__.py

@@ -10,7 +10,7 @@ For reference documentation, see https://docs.wandb.com/ref/python.
 """
 from __future__ import annotations
 
-__version__ = "0.19.1rc1"
+__version__ = "0.19.2"
 
 
 from wandb.errors import Error
@@ -204,12 +204,6 @@ if "dev" in __version__:
         "false",
     )
 
-    # Enable new features in dev versions.
-    os.environ["WANDB_X_SHOW_OPERATION_STATS"] = os.environ.get(
-        "WANDB_X_SHOW_OPERATION_STATS",
-        "true",
-    )
-
 _sentry = _Sentry()
 _sentry.setup()
 

wandb/__init__.pyi

@@ -103,7 +103,7 @@ if TYPE_CHECKING:
     import wandb
     from wandb.plot import CustomChart
 
-__version__: str = "0.19.1rc1"
+__version__: str = "0.19.2"
 
 run: Run | None
 config: wandb_config.Config
@@ -114,9 +114,7 @@ _sentry: Sentry
 api: InternalApi
 patched: Dict[str, List[Callable]]
 
-def setup(
-    settings: Settings | None = None,
-) -> Optional[_WandbSetup]:
+def setup(settings: Settings | None = None) -> _WandbSetup:
     """Prepares W&B for use in the current process and its children.
 
     You can usually ignore this as it is implicitly called by `wandb.init()`.
@@ -265,7 +263,7 @@ def init(
         entity: The username or team name under which the runs will be logged.
             The entity must already exist, so ensure you’ve created your account
             or team in the UI before starting to log runs. If not specified, the
-            run will default your defualt entity. To change the default entity,
+            run will default your default entity. To change the default entity,
             go to [your settings](https://wandb.ai/settings) and update the
             "Default location to create new projects" under "Default team".
         project: The name of the project under which this run will be logged.

wandb/agents/pyagent.py

@@ -297,7 +297,7 @@ class Agent:
                 sweep_param_path, job.config
             )
             os.environ[wandb.env.SWEEP_ID] = self._sweep_id
-            wandb.sdk.wandb_setup._setup(_reset=True)
+            wandb.teardown()
 
             wandb.termlog(f"Agent Starting Run: {run_id} with config:")
             for k, v in job.config.items():

wandb/apis/importers/wandb.py

@@ -1483,7 +1483,7 @@ def _get_run_or_dummy_from_art(art: Artifact, api=None):
         run = art.logged_by()
     except ValueError as e:
         logger.warn(
-            f"Can't log artifact because run does't exist, {art=}, {run=}, {e=}"
+            f"Can't log artifact because run doesn't exist, {art=}, {run=}, {e=}"
         )
 
     if run is not None:

wandb/apis/public/files.py

@@ -233,7 +233,7 @@ class File(Attrs):
     def _server_accepts_project_id_for_delete_file(self) -> bool:
         """Returns True if the server supports deleting files with a projectId.
 
-        This check is done by utilizing GraphQL introspection in the avaiable fields on the DeleteFiles API.
+        This check is done by utilizing GraphQL introspection in the available fields on the DeleteFiles API.
         """
         query_string = """
            query ProbeDeleteFilesProjectIdInput {

wandb/apis/public/jobs.py

@@ -407,7 +407,7 @@ class QueuedRun:
                     self._run_id = item["associatedRunId"]
                     return self._run
                 except ValueError as e:
-                    print(e)
+                    wandb.termwarn(e)
             elif item:
                 wandb.termlog("Waiting for run to start")
 

wandb/apis/public/runs.py

@@ -704,7 +704,7 @@ class Run(Attrs):
             if pd:
                 lines = pd.DataFrame.from_records(lines)
             else:
-                print("Unable to load pandas, call history with pandas=False")
+                wandb.termwarn("Unable to load pandas, call history with pandas=False")
         return lines
 
     @normalize_exceptions
@@ -908,7 +908,7 @@ class Run(Attrs):
     def _server_provides_internal_id_for_project(self) -> bool:
         """Returns True if the server allows us to query the internalId field for a project.
 
-        This check is done by utilizing GraphQL introspection in the avaiable fields on the Project type.
+        This check is done by utilizing GraphQL introspection in the available fields on the Project type.
         """
         query_string = """
            query ProbeProjectInput {
@@ -924,11 +924,6 @@ class Run(Attrs):
         if self.server_provides_internal_id_field is None:
             query = gql(query_string)
             res = self.client.execute(query)
-            print(
-                "internalId"
-                in [x["name"] for x in (res.get("ProjectType", {}).get("fields", [{}]))]
-            )
-
             self.server_provides_internal_id_field = "internalId" in [
                 x["name"] for x in (res.get("ProjectType", {}).get("fields", [{}]))
             ]

wandb/apis/reports/v1/__init__.py

@@ -4,5 +4,5 @@ try:
     from wandb_workspaces.reports.v1 import *  # noqa: F403
 except ImportError:
     wandb.termerror(
-        "Failed to import wandb_workspaces.  To edit reports programatically, please install it using `pip install wandb[workspaces]`."
+        "Failed to import wandb_workspaces.  To edit reports programmatically, please install it using `pip install wandb[workspaces]`."
     )

wandb/apis/reports/v2/__init__.py

@@ -4,5 +4,5 @@ try:
     from wandb_workspaces.reports.v2 import *  # noqa: F403
 except ImportError:
     wandb.termerror(
-        "Failed to import wandb_workspaces.  To edit reports programatically, please install it using `pip install wandb[workspaces]`."
+        "Failed to import wandb_workspaces.  To edit reports programmatically, please install it using `pip install wandb[workspaces]`."
     )

wandb/apis/workspaces/__init__.py

@@ -4,5 +4,5 @@ try:
     from wandb_workspaces.workspaces import *  # noqa: F403
 except ImportError:
     wandb.termerror(
-        "Failed to import wandb_workspaces. To edit workspaces programatically, please install it using `pip install wandb[workspaces]`."
+        "Failed to import wandb_workspaces. To edit workspaces programmatically, please install it using `pip install wandb[workspaces]`."
     )

wandb/cli/beta.py

@@ -12,6 +12,7 @@ import click
 
 import wandb
 from wandb.errors import UsageError, WandbCoreNotAvailableError
+from wandb.sdk.wandb_sync import _sync
 from wandb.util import get_core_path
 
 
@@ -108,7 +109,9 @@ def sync_beta(  # noqa: C901
                     continue
                 wandb_files = [p for p in d.glob("*.wandb") if p.is_file()]
                 if len(wandb_files) > 1:
-                    print(f"Multiple wandb files found in directory {d}, skipping")
+                    wandb.termwarn(
+                        f"Multiple wandb files found in directory {d}, skipping"
+                    )
                 elif len(wandb_files) == 1:
                     paths.add(d)
     else:
@@ -128,7 +131,7 @@ def sync_beta(  # noqa: C901
         for path in paths:
             wandb_synced_files = [p for p in path.glob("*.wandb.synced") if p.is_file()]
             if len(wandb_synced_files) > 1:
-                print(
+                wandb.termwarn(
                     f"Multiple wandb.synced files found in directory {path}, skipping"
                 )
             elif len(wandb_synced_files) == 1:
@@ -151,7 +154,7 @@ def sync_beta(  # noqa: C901
     if dry_run:
         return
 
-    wandb.sdk.wandb_setup.setup()
+    wandb.setup()
 
     # TODO: make it thread-safe in the Rust code
     with concurrent.futures.ProcessPoolExecutor(
@@ -162,7 +165,7 @@ def sync_beta(  # noqa: C901
             # we already know there is only one wandb file in the directory
             wandb_file = [p for p in path.glob("*.wandb") if p.is_file()][0]
             future = executor.submit(
-                wandb._sync,
+                _sync,
                 wandb_file,
                 run_id=run_id,
                 project=project,

wandb/cli/cli.py

@@ -125,7 +125,7 @@ def _get_cling_api(reset=None):
     global _api
     if reset:
         _api = None
-        wandb_sdk.wandb_setup._setup(_reset=True)
+        wandb.teardown()
     if _api is None:
         # TODO(jhr): make a settings object that is better for non runs.
         # only override the necessary setting
@@ -2437,7 +2437,7 @@ def ls(path, type):
                 per_page=1,
             )
             latest = next(versions)
-            print(
+            wandb.termlog(
                 "{:<15s}{:<15s}{:>15s} {:<20s}".format(
                     kind.type,
                     latest.updated_at,
@@ -2463,7 +2463,7 @@ def cleanup(target_size, remove_temp):
     target_size = util.from_human_size(target_size)
     cache = get_artifact_file_cache()
     reclaimed_bytes = cache.cleanup(target_size, remove_temp)
-    print(f"Reclaimed {util.to_human_size(reclaimed_bytes)} of space")
+    wandb.termlog(f"Reclaimed {util.to_human_size(reclaimed_bytes)} of space")
 
 
 @cli.command(context_settings=CONTEXT, help="Pull files from Weights & Biases")
@@ -2664,7 +2664,6 @@ Run `git clone {}` and restore from there or pass the --no-git flag.""".format(r
 def online():
     api = InternalApi()
     try:
-        api.clear_setting("disabled", persist=True)
         api.clear_setting("mode", persist=True)
     except configparser.Error:
         pass
@@ -2678,7 +2677,6 @@ def online():
 def offline():
     api = InternalApi()
     try:
-        api.set_setting("disabled", "true", persist=True)
         api.set_setting("mode", "offline", persist=True)
         click.echo(
             "W&B offline. Running your script from this directory will only write metadata locally. Use wandb disabled to completely turn off W&B."
@@ -2765,13 +2763,13 @@ def verify(host):
     reinit = False
     if host is None:
         host = api.settings("base_url")
-        print(f"Default host selected: {host}")
+        wandb.termlog(f"Default host selected: {host}")
     # if the given host does not match the default host, re-run init
     elif host != api.settings("base_url"):
         reinit = True
 
     tmp_dir = tempfile.mkdtemp()
-    print(
+    wandb.termlog(
         "Find detailed logs for this test at: {}".format(os.path.join(tmp_dir, "wandb"))
     )
     os.chdir(tmp_dir)

wandb/docker/__init__.py

@@ -62,7 +62,7 @@ def shell(cmd: List[str]) -> Optional[str]:
             .strip()
         )
     except subprocess.CalledProcessError as e:
-        print(e)
+        print(e)  # noqa: T201
         return None
 
 
@@ -140,12 +140,12 @@ def run_command_live_output(args: List[Any]) -> str:
                 break
             index = chunk.find(b"\r")
             if index != -1:
-                print(chunk.decode(), end="")
+                print(chunk.decode(), end="")  # noqa: T201
             else:
                 stdout += chunk.decode()
-                print(chunk.decode(), end="\r")
+                print(chunk.decode(), end="\r")  # noqa: T201
 
-        print(stdout)
+        print(stdout)  # noqa: T201
 
     return_code = process.wait()
     if return_code != 0:

wandb/integration/fastai/__init__.py

@@ -54,7 +54,7 @@ try:
         matplotlib.use("Agg")  # non-interactive backend (avoid tkinter issues)
     import matplotlib.pyplot as plt
 except ImportError:
-    print("Warning: matplotlib required if logging sample image predictions")
+    wandb.termwarn("matplotlib required if logging sample image predictions")
 
 
 class WandbCallback(TrackerCallback):
@@ -134,10 +134,8 @@ class WandbCallback(TrackerCallback):
             # Adapted from fast.ai "SaveModelCallback"
             current = self.get_monitor_value()
             if current is not None and self.operator(current, self.best):
-                print(
-                    "Better model found at epoch {} with {} value: {}.".format(
-                        epoch, self.monitor, current
-                    )
+                wandb.termlog(
+                    f"Better model found at epoch {epoch} with {self.monitor} value: {current}."
                 )
                 self.best = current
 
@@ -173,7 +171,7 @@ class WandbCallback(TrackerCallback):
             if self.model_path.is_file():
                 with self.model_path.open("rb") as model_file:
                     self.learn.load(model_file, purge=False)
-                    print(f"Loaded best saved model from {self.model_path}")
+                    wandb.termlog(f"Loaded best saved model from {self.model_path}")
 
     def _wandb_log_predictions(self) -> None:
         """Log prediction samples."""

wandb/integration/keras/keras.py

@@ -509,7 +509,9 @@ class WandbCallback(tf.keras.callbacks.Callback):
 
         # From Keras
         if mode not in ["auto", "min", "max"]:
-            print(f"WandbCallback mode {mode} is unknown, fallback to auto mode.")
+            wandb.termwarn(
+                f"WandbCallback mode {mode} is unknown, fallback to auto mode."
+            )
             mode = "auto"
 
         if mode == "min":
@@ -632,7 +634,7 @@ class WandbCallback(tf.keras.callbacks.Callback):
                 )
                 wandb.run.summary["{}{}".format(self.log_best_prefix, "epoch")] = epoch
                 if self.verbose and not self.save_model:
-                    print(
+                    wandb.termlog(
                         f"Epoch {epoch:05d}: {self.monitor} improved from {self.best:.5f} to {self.current:.5f}"
                     )
             if self.save_model:
@@ -1003,7 +1005,7 @@ class WandbCallback(tf.keras.callbacks.Callback):
         if wandb.run.disabled:
             return
         if self.verbose > 0:
-            print(
+            wandb.termlog(
                 f"Epoch {epoch:05d}: {self.monitor} improved from {self.best:.5f} to {self.current:.5f}, "
                 f"saving model to {self.filepath}"
             )

wandb/integration/metaflow/metaflow.py

@@ -73,8 +73,8 @@ try:
             wandb.termlog(f"Logging artifact: {name} ({type(data)})")
 
 except ImportError:
-    print(
-        "Warning: `pandas` not installed >> @wandb_log(datasets=True) may not auto log your dataset!"
+    wandb.termwarn(
+        "`pandas` not installed >> @wandb_log(datasets=True) may not auto log your dataset!"
     )
 
 try:
@@ -119,8 +119,8 @@ try:
             wandb.termlog(f"Logging artifact: {name} ({type(data)})")
 
 except ImportError:
-    print(
-        "Warning: `pytorch` not installed >> @wandb_log(models=True) may not auto log your model!"
+    wandb.termwarn(
+        "`pytorch` not installed >> @wandb_log(models=True) may not auto log your model!"
     )
 
 try:
@@ -164,8 +164,8 @@ try:
             wandb.termlog(f"Logging artifact: {name} ({type(data)})")
 
 except ImportError:
-    print(
-        "Warning: `sklearn` not installed >> @wandb_log(models=True) may not auto log your model!"
+    wandb.termwarn(
+        "`sklearn` not installed >> @wandb_log(models=True) may not auto log your model!"
     )
 
 
@@ -245,7 +245,7 @@ def wandb_use(name: str, data, *args, **kwargs):
     try:
         return _wandb_use(name, data, *args, **kwargs)
     except wandb.CommError:
-        print(
+        wandb.termwarn(
             f"This artifact ({name}, {type(data)}) does not exist in the wandb datastore!"
             f"If you created an instance inline (e.g. sklearn.ensemble.RandomForestClassifier), then you can safely ignore this"
             f"Otherwise you may want to check your internet connection!"

wandb/integration/prodigy/prodigy.py

@@ -237,11 +237,7 @@ def create_table(data):
                         im = Image.open(urllib.request.urlopen(document["image"]))
                         document["image_visual"] = wandb.Image(im)
                     except urllib.error.URLError:
-                        print(
-                            "Warning: Image URL "
-                            + str(document["image"])
-                            + " is invalid."
-                        )
+                        wandb.termwarn(f"Image URL {document['image']} is invalid.")
                         document["image_visual"] = None
                 elif isbase64:
                     # is base64 uri
@@ -252,11 +248,7 @@ def create_table(data):
                         im = Image.open(buf)
                         document["image_visual"] = wandb.Image(im)
                     except base64.binascii.Error:
-                        print(
-                            "Warning: Base64 string "
-                            + str(document["image"])
-                            + " is invalid."
-                        )
+                        wandb.termwarn(f"Base64 string {document['image']} is invalid.")
                         document["image_visual"] = None
                 else:
                     # is data path
@@ -296,4 +288,4 @@ def upload_dataset(dataset_name):
         standardize(data[i], schema, array_dict_types)
     table = create_table(data)
     wandb.log({dataset_name: table})
-    print("Prodigy dataset `" + dataset_name + "` uploaded.")
+    wandb.termlog(f"Prodigy dataset `{dataset_name}` uploaded.")

wandb/integration/sagemaker/__init__.py

@@ -1,12 +1,14 @@
 """wandb integration sagemaker module."""
 
 from .auth import sagemaker_auth
-from .config import parse_sm_config
-from .resources import parse_sm_resources, parse_sm_secrets
+from .config import is_using_sagemaker, parse_sm_config
+from .resources import parse_sm_secrets, set_global_settings, set_run_id
 
 __all__ = [
     "sagemaker_auth",
+    "is_using_sagemaker",
     "parse_sm_config",
     "parse_sm_secrets",
-    "parse_sm_resources",
+    "set_global_settings",
+    "set_run_id",
 ]

wandb/integration/sagemaker/config.py

@@ -1,13 +1,23 @@
+from __future__ import annotations
+
 import json
 import os
 import re
 import warnings
-from typing import Any, Dict
+from typing import Any
 
 from . import files as sm_files
 
 
-def parse_sm_config() -> Dict[str, Any]:
+def is_using_sagemaker() -> bool:
+    """Returns whether we're in a SageMaker environment."""
+    return (
+        os.path.exists(sm_files.SM_PARAM_CONFIG)  #
+        or "SM_TRAINING_ENV" in os.environ
+    )
+
+
+def parse_sm_config() -> dict[str, Any]:
     """Parses SageMaker configuration.
 
     Returns:
@@ -23,9 +33,7 @@ def parse_sm_config() -> Dict[str, Any]:
     """
     conf = {}
 
-    if os.path.exists(sm_files.SM_PARAM_CONFIG) and os.path.exists(
-        sm_files.SM_RESOURCE_CONFIG
-    ):
+    if os.path.exists(sm_files.SM_PARAM_CONFIG):
         conf["sagemaker_training_job_name"] = os.getenv("TRAINING_JOB_NAME")
 
         # Hyperparameter searches quote configs...
@@ -38,12 +46,13 @@ def parse_sm_config() -> Dict[str, Any]:
                     cast = float(cast)
                 conf[key] = cast
 
-    if "SM_TRAINING_ENV" in os.environ:
+    if env := os.environ.get("SM_TRAINING_ENV"):
         try:
-            conf = {**conf, **json.loads(os.environ["SM_TRAINING_ENV"])}
+            conf.update(json.loads(env))
         except json.JSONDecodeError:
             warnings.warn(
-                "Failed to parse SM_TRAINING_ENV not valid JSON string", stacklevel=2
+                "Failed to parse SM_TRAINING_ENV not valid JSON string",
+                stacklevel=2,
             )
 
     return conf

wandb/integration/sagemaker/files.py

@@ -1,3 +1,2 @@
 SM_PARAM_CONFIG = "/opt/ml/input/config/hyperparameters.json"
-SM_RESOURCE_CONFIG = "/opt/ml/input/config/resourceconfig.json"
 SM_SECRETS = "secrets.env"

wandb/integration/sagemaker/resources.py

@@ -1,13 +1,58 @@
+from __future__ import annotations
+
 import os
 import secrets
 import socket
 import string
-from typing import Dict, Tuple
 
+import wandb
+
+from . import config
 from . import files as sm_files
 
 
-def parse_sm_secrets() -> Dict[str, str]:
+def set_run_id(run_settings: wandb.Settings) -> bool:
+    """Set a run ID and group when using SageMaker.
+
+    Returns whether the ID and group were updated.
+    """
+    # Added in https://github.com/wandb/wandb/pull/3290.
+    #
+    # Prevents SageMaker from overriding the run ID configured
+    # in environment variables. Note, however, that it will still
+    # override a run ID passed explicitly to `wandb.init()`.
+    if os.getenv("WANDB_RUN_ID"):
+        return False
+
+    run_group = os.getenv("TRAINING_JOB_NAME")
+    if not run_group:
+        return False
+
+    alphanumeric = string.ascii_lowercase + string.digits
+    random = "".join(secrets.choice(alphanumeric) for _ in range(6))
+
+    host = os.getenv("CURRENT_HOST", socket.gethostname())
+
+    run_settings.run_id = f"{run_group}-{random}-{host}"
+    run_settings.run_group = run_group
+    return True
+
+
+def set_global_settings(settings: wandb.Settings) -> None:
+    """Set global W&B settings based on the SageMaker environment."""
+    if env := parse_sm_secrets():
+        settings.update_from_env_vars(env)
+
+    # The SageMaker config may contain an API key, in which case it
+    # takes precedence over the value in the secrets. It's unclear
+    # whether this is by design, or by accident; we keep it for
+    # backward compatibility for now.
+    sm_config = config.parse_sm_config()
+    if api_key := sm_config.get("wandb_api_key"):
+        settings.api_key = api_key
+
+
+def parse_sm_secrets() -> dict[str, str]:
     """We read our api_key from secrets.env in SageMaker."""
     env_dict = dict()
     # Set secret variables
@@ -16,19 +61,3 @@ def parse_sm_secrets() -> Dict[str, str]:
             key, val = line.strip().split("=", 1)
             env_dict[key] = val
     return env_dict
-
-
-def parse_sm_resources() -> Tuple[Dict[str, str], Dict[str, str]]:
-    run_dict = dict()
-    run_id = os.getenv("TRAINING_JOB_NAME")
-
-    if run_id and os.getenv("WANDB_RUN_ID") is None:
-        suffix = "".join(
-            secrets.choice(string.ascii_lowercase + string.digits) for _ in range(6)
-        )
-        run_dict["run_id"] = "-".join(
-            [run_id, suffix, os.getenv("CURRENT_HOST", socket.gethostname())]
-        )
-    run_dict["run_group"] = os.getenv("TRAINING_JOB_NAME")
-    env_dict = parse_sm_secrets()
-    return run_dict, env_dict

wandb/integration/torch/wandb_torch.py

@@ -441,7 +441,7 @@ class TorchGraph(wandb.data_types.Graph):
         decoder.weight encoder
         decoder.bias decoder
         """
-        # TODO: We're currently not using this, but I left it here incase we want to resurrect! - CVP
+        # TODO: We're currently not using this, but I left it here in case we want to resurrect! - CVP
         torch = util.get_module("torch", "Could not import torch")
 
         module_nodes_by_hash = {id(n): n for n in module_graph.nodes}