From 3812629e8dde63edd4ca5ba94e93d919d50a5d66 Mon Sep 17 00:00:00 2001
From: aaa123git <wandz19@mails.tsinghua.edu.cn>
Date: Thu, 20 May 2021 18:44:47 +0800
Subject: [PATCH] update get_reward (#196)
* update get_reward
* update doc
* update doc
---
convlab2/dialog_agent/env.py | 7 +------
convlab2/dialog_agent/session.py | 2 +-
convlab2/evaluator/evaluator.py | 4 ++++
convlab2/evaluator/multiwoz_eval.py | 9 +++++++++
4 files changed, 15 insertions(+), 7 deletions(-)
diff --git a/convlab2/dialog_agent/env.py b/convlab2/dialog_agent/env.py
index e26f489..8cb5616 100755
--- a/convlab2/dialog_agent/env.py
+++ b/convlab2/dialog_agent/env.py
@@ -33,12 +33,7 @@ class Environment():
state = self.sys_dst.update(dialog_act)
if self.evaluator:
- if self.evaluator.task_success():
- reward = 40
- elif self.evaluator.cur_domain and self.evaluator.domain_success(self.evaluator.cur_domain):
- reward = 5
- else:
- reward = -1
+ reward = self.evaluator.get_reward()
else:
reward = self.usr.get_reward()
terminated = self.usr.is_terminated()
diff --git a/convlab2/dialog_agent/session.py b/convlab2/dialog_agent/session.py
index ffe70de..64fe8df 100755
--- a/convlab2/dialog_agent/session.py
+++ b/convlab2/dialog_agent/session.py
@@ -126,7 +126,7 @@ class BiSession(Session):
# print('inform prec. {} rec. {} F1 {}'.format(prec, rec, f1))
# print('book rate {}'.format(self.evaluator.book_rate()))
# print('task success {}'.format(self.evaluator.task_success()))
- reward = self.user_agent.get_reward()
+ reward = self.user_agent.get_reward() if self.evaluator is None else self.evaluator.get_reward()
sys_response = self.next_response(user_response)
self.dialog_history.append([self.user_agent.name, user_response])
self.dialog_history.append([self.sys_agent.name, sys_response])
diff --git a/convlab2/evaluator/evaluator.py b/convlab2/evaluator/evaluator.py
index 89a37b3..7857023 100755
--- a/convlab2/evaluator/evaluator.py
+++ b/convlab2/evaluator/evaluator.py
@@ -55,3 +55,7 @@ class Evaluator(object):
def final_goal_analyze(self):
"""judge whether the final goal satisfies the database constraints"""
raise NotImplementedError
+
+ def get_reward(self):
+ """returns a reward, which is used for RL training."""
+ raise NotImplementedError
diff --git a/convlab2/evaluator/multiwoz_eval.py b/convlab2/evaluator/multiwoz_eval.py
index e7303c5..11dd0dd 100755
--- a/convlab2/evaluator/multiwoz_eval.py
+++ b/convlab2/evaluator/multiwoz_eval.py
@@ -413,3 +413,12 @@ class MultiWozEvaluator(Evaluator):
return 1
else:
return match / (match + mismatch)
+
+ def get_reward(self):
+ if self.task_success():
+ reward = 40
+ elif self.cur_domain and self.domain_success(self.cur_domain):
+ reward = 5
+ else:
+ reward = -1
+ return reward
--
GitLab