Baseline results from Shakespeare Paraphraser added

baseline/baseline.py

@@ -0,0 +1,45 @@
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
+import re
+import argparse
+
+
+def main():
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--filename', dest='filename', type=str, help='Name of input file')
+    parser.add_argument('--outfile', dest='outfile', type=str, help='Name of output file')
+
+    args = parser.parse_args()
+    filename = args.filename
+    outfile = args.outfile
+
+    device = "cuda:0" if torch.cuda.is_available() else "cpu"
+    print("Device is ", device)
+
+    tokenizer = AutoTokenizer.from_pretrained("filco306/gpt2-shakespeare-paraphraser")
+    model = AutoModelForCausalLM.from_pretrained("filco306/gpt2-shakespeare-paraphraser", pad_token_id=tokenizer.eos_token_id)
+    model.to(device)
+
+    base_input_path = "./shakespeare/sparknotes/merged/"
+    base_output_path = "./output/"
+
+    self.generate(base_input_path + filename, base_output_path + outfile, tokenizer, model, device)
+
+
+def generate(input_path, output_path, tokenizer, model, device):
+
+    input = None
+    for filename in files:
+        f = open(input_path, "r")
+        input = f.read()
+
+    output_file = open(output_path, "w")
+    torch.manual_seed(0)
+    sentences = re.split('\n', input)
+    for sentence in sentences:
+        input_ids = tokenizer.encode(sentence, return_tensors='pt').to(device)
+        greedy_output = model.generate(input_ids, max_length=input_ids.shape[1]*1.5, early_stopping = True, top_p=0.90)
+        output_file.write('\n')
+        output_file.write(tokenizer.decode(greedy_output[0], skip_special_tokens=True))
+

baseline/run.sh

@@ -0,0 +1,17 @@
+python3 baseline.py --filename antony-and-cleopatra_original.snt.aligned --outfile antony-and-cleopatra_translated.txt
+python3 baseline.py --filename asyoulikeit_original.snt.aligned --outfile asyoulikeit_translated.txt
+python3 baseline.py --filename errors_original.snt.aligned --outfile errors_translated.txt
+python3 baseline.py --filename hamlet_original.snt.aligned --outfile hamlet_translated.txt
+python3 baseline.py --filename henryv_original.snt.aligned --outfile henryv_translated.txt
+python3 baseline.py --filename juliuscaesar_original.snt.aligned --outfile juliuscaeser_translated.txt
+python3 baseline.py --filename lear_original.snt.aligned --outfile lear_translated.txt
+python3 baseline.py --filename macbeth_original.snt.aligned --outfile macbeth_translated.txt
+python3 baseline.py --filename merchant_original.snt.aligned --outfile merchant_translated.txt
+python3 baseline.py --filename msnd_original.snt.aligned --outfile msnd_translated.txt
+python3 baseline.py --filename muchado_original.snt.aligned --outfile muchado_translated.txt
+python3 baseline.py --filename othello_original.snt.aligned --outfile othello_translated.txt
+python3 baseline.py --filename richardiii_original.snt.aligned --outfile richardiii_translated.txt
+python3 baseline.py --filename romeojuliet_original.snt.aligned --outfile romeojuliet_translated.txt
+python3 baseline.py --filename shrew_original.snt.aligned --outfile shrew_translated.txt
+python3 baseline.py --filename tempest_original.snt.aligned --outfile tempest_translated.txt
+python3 baseline.py --filename twelfthnight_original.snt.aligned --outfile twelfthnight_translated.txt