Merge pull request #8 from asappresearch/golden_agent

Golden agent

Author: blattimer-asapp

josh_train/async_josh.py

@@ -107,7 +107,7 @@ async def step_agent(self):
                             print(f'🌟 Got reward')
                         collapse_root_to = unfinished_leaves[idx]
                         rewards_to_delete = copy.deepcopy(rw_to_delete)
-                        if len(self.rewards)==1:
+                        if len(self.rewards)-len(rewards_to_delete)==0:
                             self.golden_agent = turn.agent
 
                 if count+1 == self.max_turn_tries and not turn_finished:

josh_train/josh.py

@@ -198,7 +198,7 @@ def step_agent(self):
                             print(f'🌟 Got reward')
                         collapse_root_to = unfinished_leaves[idx]
                         rewards_to_delete = copy.deepcopy(rw_to_delete)
-                        if len(self.rewards)==1:
+                        if len(self.rewards)-len(rewards_to_delete)==0:
                             self.golden_agent = turn.agent
 
                 if count+1 == self.max_turn_tries:

setup.py

@@ -25,7 +25,7 @@ def get_requires(requires_filename: str) -> List[str]:
     install_requires=get_requires("requirements.txt"),
     include_package_data=True,
     setup_requires=["setuptools_scm"],
-    version="0.1.3",
+    version="0.1.4",
     classifiers=[
         "Programming Language :: Python :: 3",
         "License :: OSI Approved :: MIT License",