Update app.py
Browse files
app.py
CHANGED
@@ -20,8 +20,20 @@ def extract_sentences_from_excel(file):
|
|
20 |
return sentences
|
21 |
|
22 |
def save_to_json(data, filename="synthetic_data.json"):
|
23 |
-
with open(filename, mode='
|
24 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
|
26 |
def generate(file, prompt, temperature, max_new_tokens, top_p, repetition_penalty):
|
27 |
sentences = extract_sentences_from_excel(file)
|
|
|
20 |
return sentences
|
21 |
|
22 |
def save_to_json(data, filename="synthetic_data.json"):
|
23 |
+
with open(filename, mode='w', encoding='utf-8') as file:
|
24 |
+
json_data = []
|
25 |
+
for item in data:
|
26 |
+
generated_sentences = []
|
27 |
+
confidence_scores = []
|
28 |
+
for match in re.finditer(r"{'generated_sentence': '(.+)', 'confidence_score': (\d+\.\d+)}", item['generated_data']):
|
29 |
+
generated_sentences.append(match.group(1))
|
30 |
+
confidence_scores.append(float(match.group(2)))
|
31 |
+
json_data.append({
|
32 |
+
'original_sentence': item['original_sentence'],
|
33 |
+
'generated_sentences': generated_sentences,
|
34 |
+
'confidence_scores': confidence_scores
|
35 |
+
})
|
36 |
+
json.dump(json_data, file, indent=4, ensure_ascii=False)
|
37 |
|
38 |
def generate(file, prompt, temperature, max_new_tokens, top_p, repetition_penalty):
|
39 |
sentences = extract_sentences_from_excel(file)
|