From e3a110f03e2d03dd550bf4e49a45616240db2752 Mon Sep 17 00:00:00 2001
From: Franck Dary <franck.dary@lis-lab.fr>
Date: Tue, 6 Jul 2021 13:43:42 +0200
Subject: [PATCH] Corrected bug where system wanted to back at each end of
sentence
---
Train.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/Train.py b/Train.py
index 0724dbd..e56c47e 100644
--- a/Train.py
+++ b/Train.py
@@ -254,9 +254,9 @@ def trainModelRl(debug, networkName, modelDir, filename, nbIter, batchSize, devF
reward = torch.FloatTensor([reward_]).to(getDevice())
newState = None
+ toState = strategy[action.name][1] if action.name in strategy else -1
if appliable :
applyTransition(strategy, sentence, action, reward_)
- toState = sentence.state
newState = policy_net.extractFeatures(dicts, sentence).to(getDevice())
else:
count+=1
--
GitLab