monai-weekly 1.5.dev2443__tar.gz → 1.5.dev2444__tar.gz

{monai_weekly-1.5.dev2443 → monai_weekly-1.5.dev2444}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: monai-weekly
-Version: 1.5.dev2443
+Version: 1.5.dev2444
 Summary: AI Toolkit for Healthcare Imaging
 Home-page: https://monai.io/
 Author: MONAI Consortium

{monai_weekly-1.5.dev2443 → monai_weekly-1.5.dev2444}/monai/__init__.py

@@ -136,4 +136,4 @@ except BaseException:
 
     if MONAIEnvVars.debug():
         raise
-__commit_id__ = "82298adb01c27a6cbf0bf10a9916b09b41966bb0"
+__commit_id__ = "c1ceea3d4cbb0781eae4e209b80fe651a776fed2"

{monai_weekly-1.5.dev2443 → monai_weekly-1.5.dev2444}/monai/_version.py

@@ -8,11 +8,11 @@ import json
 
 version_json = '''
 {
- "date": "2024-10-27T02:29:25+0000",
+ "date": "2024-11-03T02:29:11+0000",
  "dirty": false,
  "error": null,
- "full-revisionid": "d508ba780862a6459326c812e09188838c25bdc5",
- "version": "1.5.dev2443"
+ "full-revisionid": "c70fbd8ff919cabaacfabcbdbf28aa435ae622f9",
+ "version": "1.5.dev2444"
 }
 '''  # END VERSION_JSON
 

{monai_weekly-1.5.dev2443 → monai_weekly-1.5.dev2444}/monai/handlers/mlflow_handler.py

@@ -22,7 +22,7 @@ import torch
 from torch.utils.data import Dataset
 
 from monai.apps.utils import get_logger
-from monai.utils import CommonKeys, IgniteInfo, ensure_tuple, min_version, optional_import
+from monai.utils import CommonKeys, IgniteInfo, ensure_tuple, flatten_dict, min_version, optional_import
 
 Events, _ = optional_import("ignite.engine", IgniteInfo.OPT_IMPORT_VERSION, min_version, "Events")
 mlflow, _ = optional_import("mlflow", descriptor="Please install mlflow before using MLFlowHandler.")
@@ -303,7 +303,9 @@ class MLFlowHandler:
 
         run_id = self.cur_run.info.run_id
         timestamp = int(time.time() * 1000)
-        metrics_arr = [mlflow.entities.Metric(key, value, timestamp, step or 0) for key, value in metrics.items()]
+        metrics_arr = [
+            mlflow.entities.Metric(key, value, timestamp, step or 0) for key, value in flatten_dict(metrics).items()
+        ]
         self.client.log_batch(run_id=run_id, metrics=metrics_arr, params=[], tags=[])
 
     def _parse_artifacts(self):

{monai_weekly-1.5.dev2443 → monai_weekly-1.5.dev2444}/monai/handlers/stats_handler.py

@@ -19,7 +19,7 @@ from typing import TYPE_CHECKING, Any
 import torch
 
 from monai.apps import get_logger
-from monai.utils import IgniteInfo, is_scalar, min_version, optional_import
+from monai.utils import IgniteInfo, flatten_dict, is_scalar, min_version, optional_import
 
 Events, _ = optional_import("ignite.engine", IgniteInfo.OPT_IMPORT_VERSION, min_version, "Events")
 if TYPE_CHECKING:
@@ -211,8 +211,7 @@ class StatsHandler:
 
         """
         current_epoch = self.global_epoch_transform(engine.state.epoch)
-
-        prints_dict = engine.state.metrics
+        prints_dict = flatten_dict(engine.state.metrics)
         if prints_dict is not None and len(prints_dict) > 0:
             out_str = f"Epoch[{current_epoch}] Metrics -- "
             for name in sorted(prints_dict):

{monai_weekly-1.5.dev2443 → monai_weekly-1.5.dev2444}/monai/networks/utils.py

@@ -16,6 +16,7 @@ from __future__ import annotations
 
 import io
 import re
+import tempfile
 import warnings
 from collections import OrderedDict
 from collections.abc import Callable, Mapping, Sequence
@@ -688,16 +689,17 @@ def convert_to_onnx(
             onnx_inputs = (inputs,)
         else:
             onnx_inputs = tuple(inputs)
-
+        temp_file = None
         if filename is None:
-            f = io.BytesIO()
+            temp_file = tempfile.NamedTemporaryFile()
+            f = temp_file.name
         else:
             f = filename
 
         torch.onnx.export(
             mode_to_export,
             onnx_inputs,
-            f=f,  # type: ignore[arg-type]
+            f=f,
             input_names=input_names,
             output_names=output_names,
             dynamic_axes=dynamic_axes,
@@ -705,10 +707,7 @@ def convert_to_onnx(
             do_constant_folding=do_constant_folding,
             **torch_versioned_kwargs,
         )
-        if filename is None:
-            onnx_model = onnx.load_model_from_string(f.getvalue())
-        else:
-            onnx_model = onnx.load(filename)
+        onnx_model = onnx.load(f)
 
     if do_constant_folding and polygraphy_imported:
         from polygraphy.backend.onnx.loader import fold_constants

{monai_weekly-1.5.dev2443 → monai_weekly-1.5.dev2444}/monai/utils/__init__.py

@@ -78,6 +78,7 @@ from .misc import (
     ensure_tuple_size,
     fall_back_tuple,
     first,
+    flatten_dict,
     get_seed,
     has_option,
     is_immutable,

{monai_weekly-1.5.dev2443 → monai_weekly-1.5.dev2444}/monai/utils/misc.py

@@ -916,3 +916,16 @@ def unsqueeze_right(arr: NT, ndim: int) -> NT:
 def unsqueeze_left(arr: NT, ndim: int) -> NT:
     """Prepend 1-sized dimensions to `arr` to create a result with `ndim` dimensions."""
     return arr[(None,) * (ndim - arr.ndim)]
+
+
+def flatten_dict(metrics: dict[str, Any]) -> dict[str, Any]:
+    """
+    Flatten the nested dictionary to a flat dictionary.
+    """
+    result = {}
+    for key, value in metrics.items():
+        if isinstance(value, dict):
+            result.update(flatten_dict(value))
+        else:
+            result[key] = value
+    return result

{monai_weekly-1.5.dev2443 → monai_weekly-1.5.dev2444}/monai_weekly.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: monai-weekly
-Version: 1.5.dev2443
+Version: 1.5.dev2444
 Summary: AI Toolkit for Healthcare Imaging
 Home-page: https://monai.io/
 Author: MONAI Consortium

{monai_weekly-1.5.dev2443 → monai_weekly-1.5.dev2444}/tests/test_handler_mlflow.py

@@ -122,6 +122,11 @@ class TestHandlerMLFlow(unittest.TestCase):
             def _update_metric(engine):
                 current_metric = engine.state.metrics.get("acc", 0.1)
                 engine.state.metrics["acc"] = current_metric + 0.1
+                # log nested metrics
+                engine.state.metrics["acc_per_label"] = {
+                    "label_0": current_metric + 0.1,
+                    "label_1": current_metric + 0.2,
+                }
                 engine.state.test = current_metric
 
             # set up testing handler
@@ -138,10 +143,12 @@ class TestHandlerMLFlow(unittest.TestCase):
                 state_attributes=["test"],
                 experiment_param=experiment_param,
                 artifacts=[artifact_path],
-                close_on_complete=True,
+                close_on_complete=False,
             )
             handler.attach(engine)
             engine.run(range(3), max_epochs=2)
+            cur_run = handler.client.get_run(handler.cur_run.info.run_id)
+            self.assertTrue("label_0" in cur_run.data.metrics.keys())
             handler.close()
             # check logging output
             self.assertTrue(len(glob.glob(test_path)) > 0)