learning-loop-node 0.19.0__py3-none-any.whl → 0.20.1__py3-none-any.whl

learning_loop_node/trainer/trainer_logic.py

@@ -79,9 +79,12 @@ class TrainerLogic(TrainerLogicGeneric):
 
         except TrainingError:
             logging.exception('Exception in trainer_logic._train')
-            await self.executor.stop_and_wait()
             self.training.training_state = previous_state
             raise
+        finally:
+            # shield so the subprocess is stopped even when the task is cancelled (e.g. on abort())
+            if self._executor and self._executor.is_running():
+                await asyncio.shield(self.executor.stop_and_wait())
 
     async def _do_detections(self) -> None:
         context = self.training.context
@@ -147,20 +150,12 @@ class TrainerLogic(TrainerLogicGeneric):
     async def stop(self) -> None:
         """If executor is running, stop it. Else cancel training task."""
         print('===============> stop received in trainer_logic.', flush=True)
-
         if not self.training_active:
             return
         if self._executor and self._executor.is_running():
             await self.executor.stop_and_wait()
-        elif self.training_task:
-            logging.info('cancelling training task')
-            if self.training_task.cancel():
-                try:
-                    await self.training_task
-                except asyncio.CancelledError:
-                    pass
-                logging.info('cancelled training task')
-                self._may_restart()
+            return
+        await super().stop()
 
     # ---------------------------------------- ABSTRACT METHODS ----------------------------------------
 

learning_loop_node/trainer/trainer_logic_generic.py

@@ -470,18 +470,23 @@ class TrainerLogicGeneric(ABC):
         await self.stop()
 
     async def stop(self):
-        """Stops the training process by canceling training task.
+        """Stop the training. Default is to abort.
+        """
+        await self.abort()
+
+    async def abort(self):
+        """Abort the training: cancel the task immediately and discards in-progress model.
         """
         if not self.training_active:
             return
         if self.training_task:
-            logger.info('cancelling training task')
+            logger.info('aborting training task')
             if self.training_task.cancel():
                 try:
                     await self.training_task
                 except asyncio.CancelledError:
                     pass
-                logger.info('cancelled training task')
+                logger.info('aborted training task')
                 self._may_restart()
 
     def _may_restart(self) -> None:

learning_loop_node/trainer/trainer_node.py

@@ -76,6 +76,15 @@ class TrainerNode(Node):
                 self.log.exception('error in stop_training. Exception:')
             return True
 
+        @sio_client.event
+        async def abort_training():
+            self.log.info('abort_training received. Current state : %s', self.trainer_logic.state)
+            try:
+                await self.trainer_logic.abort()
+            except Exception:
+                self.log.exception('error in abort_training. Exception:')
+            return True
+
     async def send_status(self):
         if not self.sio_client or not self.sio_client.connected:
             self.log.debug('cannot send status - not connected to the Learning Loop')

{learning_loop_node-0.19.0.dist-info → learning_loop_node-0.20.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: learning_loop_node
-Version: 0.19.0
+Version: 0.20.1
 Summary: Python Library for Nodes which connect to the Zauberzeug Learning Loop
 Author-email: Zauberzeug GmbH <info@zauberzeug.com>
 License: MIT

{learning_loop_node-0.19.0.dist-info → learning_loop_node-0.20.1.dist-info}/RECORD

@@ -50,12 +50,12 @@ learning_loop_node/trainer/exceptions.py,sha256=vbuoE6kssLQuA8zd3LiDHmZglP6E2IJJ
 learning_loop_node/trainer/executor.py,sha256=XQ1TwgOF2pQv7bgzSaljDWG4I1ySjYXL8r9c_p9MX1Q,3947
 learning_loop_node/trainer/io_helpers.py,sha256=zWl0-HLP8zxOG3ybm76vXgEozULj_1QRl95jkrF126k,7255
 learning_loop_node/trainer/test_executor.py,sha256=6BVGDN_6f5GEMMEvDLSG1yzMybSvgXaP5uYpSfsVPP0,2224
-learning_loop_node/trainer/trainer_logic.py,sha256=eK-01qZzi10UjLMCQX8vy5eW2FoghPj3rzzDC-s3Si4,8792
-learning_loop_node/trainer/trainer_logic_generic.py,sha256=KcHmXr-Hp8_Wuejzj8odY6sRPqi6aw1SEXv3YlbjM98,27057
-learning_loop_node/trainer/trainer_node.py,sha256=tsAMzJewdS7Bi_1b9FwG0d2lGlv2lY37pgOLWr0bP_I,4582
+learning_loop_node/trainer/trainer_logic.py,sha256=BgkeDEVaRVuRETYMt2rKe5cspExTgnO7LXNVYVbIcsw,8679
+learning_loop_node/trainer/trainer_logic_generic.py,sha256=IRP68ZGpnOX4OS03OwB7HmTUvk6nbBRmCjMCVupGE_Q,27196
+learning_loop_node/trainer/trainer_node.py,sha256=ZnydfHJ2ehAhb821Xke_09qthCE3vDf7iqeYUrliSSo,4938
 learning_loop_node/trainer/rest/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 learning_loop_node/trainer/rest/backdoor_controls.py,sha256=ZnK8ypY5r_q0-YZbtaOxhQThzuZvMsQHM5gJGESd_dE,5131
-learning_loop_node-0.19.0.dist-info/METADATA,sha256=XrZLdSNokLlWJvcMmORSSxkx3HhYy9t9bCyjoK-a0Bk,15383
-learning_loop_node-0.19.0.dist-info/WHEEL,sha256=aeYiig01lYGDzBgS8HxWXOg3uV61G9ijOsup-k9o1sk,91
-learning_loop_node-0.19.0.dist-info/top_level.txt,sha256=nS5qiFc4stVF_VXVCVJxZnrqVOv3IGX5-_E3sLby6OE,19
-learning_loop_node-0.19.0.dist-info/RECORD,,
+learning_loop_node-0.20.1.dist-info/METADATA,sha256=8dCz3CJ2-uiWstXKUzKrRirUdukjQP3nbqDLdcTcIhU,15383
+learning_loop_node-0.20.1.dist-info/WHEEL,sha256=K260EYznzXsJYBQGqmI8VTxEdiZYNvDZwW9cBh9-_MA,91
+learning_loop_node-0.20.1.dist-info/top_level.txt,sha256=nS5qiFc4stVF_VXVCVJxZnrqVOv3IGX5-_E3sLby6OE,19
+learning_loop_node-0.20.1.dist-info/RECORD,,

{learning_loop_node-0.19.0.dist-info → learning_loop_node-0.20.1.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (82.0.1)
+Generator: setuptools (83.0.0)
 Root-Is-Purelib: true
 Tag: py3-none-any