Spaces:
Running
Running
Update src/paraphrase/Paraphrase.py
Browse files- src/paraphrase/Paraphrase.py +17 -23
src/paraphrase/Paraphrase.py
CHANGED
|
@@ -3,35 +3,29 @@ from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
|
| 3 |
import torch
|
| 4 |
import src.exception.Exception.Exception as ExceptionCustom
|
| 5 |
|
| 6 |
-
|
| 7 |
METHOD = "PARAPHRASE"
|
| 8 |
|
| 9 |
-
|
| 10 |
tokenizer = AutoTokenizer.from_pretrained("BlackKakapo/flan-t5-base-paraphrase-ro")
|
| 11 |
model = AutoModelForSeq2SeqLM.from_pretrained("BlackKakapo/flan-t5-base-paraphrase-ro")
|
| 12 |
|
| 13 |
# device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 14 |
# model.to(device)
|
| 15 |
|
| 16 |
-
|
| 17 |
def paraphraseParaphraseMethod(requestValue : str):
|
|
|
|
|
|
|
| 18 |
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
|
| 23 |
-
|
| 24 |
-
if exception != "":
|
| 25 |
-
return "", exception
|
| 26 |
|
| 27 |
-
|
|
|
|
| 28 |
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
text = "paraphrase: " + SENTENCE
|
| 32 |
-
|
| 33 |
-
encoding = tokenizer.encode_plus(text, pad_to_max_length=True, return_tensors="pt")
|
| 34 |
-
input_ids, attention_masks = encoding["input_ids"], encoding["attention_mask"]
|
| 35 |
|
| 36 |
beam_outputs = model.generate(
|
| 37 |
input_ids=input_ids,
|
|
@@ -46,11 +40,11 @@ def paraphraseParaphraseMethod(requestValue : str):
|
|
| 46 |
num_beams=1
|
| 47 |
)
|
| 48 |
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
|
| 56 |
-
|
|
|
|
| 3 |
import torch
|
| 4 |
import src.exception.Exception.Exception as ExceptionCustom
|
| 5 |
|
|
|
|
| 6 |
METHOD = "PARAPHRASE"
|
| 7 |
|
|
|
|
| 8 |
tokenizer = AutoTokenizer.from_pretrained("BlackKakapo/flan-t5-base-paraphrase-ro")
|
| 9 |
model = AutoModelForSeq2SeqLM.from_pretrained("BlackKakapo/flan-t5-base-paraphrase-ro")
|
| 10 |
|
| 11 |
# device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 12 |
# model.to(device)
|
| 13 |
|
|
|
|
| 14 |
def paraphraseParaphraseMethod(requestValue : str):
|
| 15 |
+
exception = ""
|
| 16 |
+
result_value = ""
|
| 17 |
|
| 18 |
+
exception = ExceptionCustom.checkForException(requestValue, METHOD)
|
| 19 |
+
if exception != "":
|
| 20 |
+
return "", exception
|
| 21 |
|
| 22 |
+
tokenized_sent_list = sent_tokenize(requestValue)
|
|
|
|
|
|
|
| 23 |
|
| 24 |
+
for SENTENCE in tokenized_sent_list:
|
| 25 |
+
text = "paraphrase: " + SENTENCE
|
| 26 |
|
| 27 |
+
encoding = tokenizer.encode_plus(text, pad_to_max_length=True, return_tensors="pt")
|
| 28 |
+
input_ids, attention_masks = encoding["input_ids"], encoding["attention_mask"]
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
|
| 30 |
beam_outputs = model.generate(
|
| 31 |
input_ids=input_ids,
|
|
|
|
| 40 |
num_beams=1
|
| 41 |
)
|
| 42 |
|
| 43 |
+
for beam_output in beam_outputs:
|
| 44 |
+
text_para = tokenizer.decode(beam_output, skip_special_tokens=True, clean_up_tokenization_spaces=True)
|
| 45 |
+
|
| 46 |
+
if SENTENCE.lower().strip() != text_para.lower().strip():
|
| 47 |
+
result_value += text_para + " "
|
| 48 |
+
break
|
| 49 |
|
| 50 |
+
return result_value, ""
|