From 98913c4369faf46ad432375363216b0705f96ce6 Mon Sep 17 00:00:00 2001
From: Aman Ashray <ashray17aman@gmail.com>
Date: Tue, 25 Jun 2019 12:37:20 +1000
Subject: [PATCH 1/3] Update README.md

---
 README.md | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 5e355d4..87879e3 100644
--- a/README.md
+++ b/README.md
@@ -1,3 +1,5 @@
+Forked and corrected. working model
+
 # Simple Word-based Language Model in PyTorch
 This model is directly analagous to this [Tesnsorflow's LM](https://www.tensorflow.org/tutorials/recurrent).
 In fact, the reader is directly taken from its older version
@@ -21,4 +23,4 @@ On GTX 1080 I am getting around 7,400 wps.
 
 ## Requirements
 * Python 3 (I used Anaconda distribution)
-* PyTorch (I used 0.1.12)
\ No newline at end of file
+* PyTorch (I used 0.1.12)

From c52bbed5963f8ac8c853633f8532540423b43c45 Mon Sep 17 00:00:00 2001
From: Aman Ashray <ashray17aman@gmail.com>
Date: Tue, 25 Jun 2019 12:37:39 +1000
Subject: [PATCH 2/3] Update README.md

---
 README.md | 2 --
 1 file changed, 2 deletions(-)

diff --git a/README.md b/README.md
index 87879e3..04a7806 100644
--- a/README.md
+++ b/README.md
@@ -1,5 +1,3 @@
-Forked and corrected. working model
-
 # Simple Word-based Language Model in PyTorch
 This model is directly analagous to this [Tesnsorflow's LM](https://www.tensorflow.org/tutorials/recurrent).
 In fact, the reader is directly taken from its older version

From 9c44e01799a3b5c27127122b69e5ce0d89642823 Mon Sep 17 00:00:00 2001
From: Aman Ashray <ashray17aman@gmail.com>
Date: Tue, 25 Jun 2019 12:40:30 +1000
Subject: [PATCH 3/3] Update ptb-lm.py

---
 ptb-lm.py | 19 +++++++++++++++----
 1 file changed, 15 insertions(+), 4 deletions(-)

diff --git a/ptb-lm.py b/ptb-lm.py
index f43c754..eb49e89 100644
--- a/ptb-lm.py
+++ b/ptb-lm.py
@@ -7,9 +7,10 @@
 from lm import repackage_hidden, LM_LSTM
 import reader
 import numpy as np
+import copy
 
 parser = argparse.ArgumentParser(description='Simplest LSTM-based language model in PyTorch')
-parser.add_argument('--data', type=str, default='data',
+parser.add_argument('--data', type=str, default='data/penn-treebank/',
                     help='location of the data corpus')
 parser.add_argument('--hidden_size', type=int, default=1500,
                     help='size of word embeddings')
@@ -39,18 +40,28 @@ def run_epoch(model, data, is_train=False, lr=1.0):
   epoch_size = ((len(data) // model.batch_size) - 1) // model.num_steps
   start_time = time.time()
   hidden = model.init_hidden()
+  savehidden = copy.deepcopy(hidden)
   costs = 0.0
   iters = 0
   for step, (x, y) in enumerate(reader.ptb_iterator(data, model.batch_size, model.num_steps)):
+    #print(x,y)
+
     inputs = Variable(torch.from_numpy(x.astype(np.int64)).transpose(0, 1).contiguous()).cuda()
     model.zero_grad()
-    hidden = repackage_hidden(hidden)
+    #print(hidden)
+    #print(x,'   x ', y, ' print x and y')
+    '''
+     hidden = repackage_hidden(hidden)
+     modified - ashray17aman
+    '''
+    hidden = copy.deepcopy(savehidden)
     outputs, hidden = model(inputs, hidden)
     targets = Variable(torch.from_numpy(y.astype(np.int64)).transpose(0, 1).contiguous()).cuda()
     tt = torch.squeeze(targets.view(-1, model.batch_size * model.num_steps))
 
     loss = criterion(outputs.view(-1, model.vocab_size), tt)
-    costs += loss.data[0] * model.num_steps
+    #print(loss)
+    costs += torch.Tensor.item(loss.data) * model.num_steps
     iters += model.num_steps
 
     if is_train:
@@ -90,4 +101,4 @@ def run_epoch(model, data, is_train=False, lr=1.0):
   print('Test Perplexity: {:8.2f}'.format(run_epoch(model, test_data)))
   with open(args.save, 'wb') as f:
     torch.save(model, f)
-  print("########## Done! ##########################")
\ No newline at end of file
+  print("########## Done! ##########################")