Spaces:
Sleeping
Sleeping
add models.py
Browse files
models.py
ADDED
@@ -0,0 +1,108 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import parser
|
2 |
+
import requests
|
3 |
+
|
4 |
+
def zephyr_score(sentence):
|
5 |
+
prompt = f"""<|user|>
|
6 |
+
You are an assistant helping with paper reviews.
|
7 |
+
You will be tasked to classify sentences as 'J' or 'V'
|
8 |
+
|
9 |
+
'J' is positive or 'J' is encouraging.
|
10 |
+
'J' has a neutral tone or 'J' is professional.
|
11 |
+
'V' is overly blunt or 'V' contains excessive negativity and no constructive feedback.
|
12 |
+
'V' contains an accusatory tone or 'V' contains sweeping generalizations or 'V' contains personal attacks.
|
13 |
+
|
14 |
+
Text: "{sentence}"
|
15 |
+
|
16 |
+
Please classify this text as either 'J', 'W', or 'V'. Only output 'J', 'W', or 'V' with no additional explanation.<|endoftext|>
|
17 |
+
<|assistant|>
|
18 |
+
"""
|
19 |
+
return prompt
|
20 |
+
|
21 |
+
def zephyr_revise(sentence):
|
22 |
+
prompt = f"""<|user|>
|
23 |
+
You are an assistant that helps users revise Paper Reviews.
|
24 |
+
Paper reviews exist to provide authors of academic research papers constructive critism.
|
25 |
+
|
26 |
+
This is text found in a review.
|
27 |
+
This text was classified as 'toxic':
|
28 |
+
|
29 |
+
Text: "{sentence}"
|
30 |
+
|
31 |
+
Please revise this text such that it maintains the criticism in the original text and delivers it in a friendly but professional manner. Make minimal changes to the original text.<|endoftext|>
|
32 |
+
<|assistant|>
|
33 |
+
"""
|
34 |
+
return prompt
|
35 |
+
|
36 |
+
def query_model_score(sentence, api_key, model_id, prompt_fun):
|
37 |
+
API_URL = f"https://api-inference.huggingface.co/models/{model_id}"
|
38 |
+
headers = {"Authorization": f"Bearer {api_key}"}
|
39 |
+
prompt = prompt_fun(sentence)
|
40 |
+
def query(payload):
|
41 |
+
print(payload)
|
42 |
+
response = requests.post(API_URL, headers=headers, json=payload)
|
43 |
+
return response.json()
|
44 |
+
parameters = {"max_new_tokens" : 20, "temperature": 0.0, "return_full_text": False}
|
45 |
+
options = {"wait_for_model": True}
|
46 |
+
data = query({"inputs": f"{prompt}", "parameters": parameters, "options": options})
|
47 |
+
score = data[0]['generated_text']
|
48 |
+
if 'v' in score.lower():
|
49 |
+
return 1
|
50 |
+
else:
|
51 |
+
return 0
|
52 |
+
|
53 |
+
def query_model_revise(sentence, api_key, model_id, prompt_fun):
|
54 |
+
API_URL = f"https://api-inference.huggingface.co/models/{model_id}"
|
55 |
+
headers = {"Authorization": f"Bearer {api_key}"}
|
56 |
+
prompt = prompt_fun(sentence)
|
57 |
+
def query(payload):
|
58 |
+
response = requests.post(API_URL, headers=headers, json=payload)
|
59 |
+
return response.json()
|
60 |
+
parameters = {"max_new_tokens" : 200, "temperature": 0.0, "return_full_text": False}
|
61 |
+
options = {"wait_for_model": True}
|
62 |
+
data = query({"inputs": f"{prompt}", "parameters": parameters, "options": options})
|
63 |
+
revision = data[0]['generated_text']
|
64 |
+
return revision
|
65 |
+
|
66 |
+
def revise_review(review, api_key, model_id, highlight_color):
|
67 |
+
result = {
|
68 |
+
"success": False,
|
69 |
+
"data": {
|
70 |
+
"revision": "",
|
71 |
+
"score": "",
|
72 |
+
"sentence_count": "",
|
73 |
+
"revised_sentences": ""
|
74 |
+
},
|
75 |
+
"message": ""
|
76 |
+
}
|
77 |
+
|
78 |
+
try:
|
79 |
+
review = review.replace('"', "'")
|
80 |
+
sentences = parser.parse_sentences(review)
|
81 |
+
review_score = 0
|
82 |
+
revision_count = 0
|
83 |
+
review_revision = ""
|
84 |
+
for sentence in sentences:
|
85 |
+
if len(sentence) > 20:
|
86 |
+
score = query_model_score(sentence, api_key, model_id, zephyr_score)
|
87 |
+
if score == 0:
|
88 |
+
review_revision += " " + sentence
|
89 |
+
else:
|
90 |
+
review_score = 1
|
91 |
+
revision_count +=1
|
92 |
+
revision = query_model_revise(sentence, api_key, model_id, zephyr_revise)
|
93 |
+
revision = revision.strip().strip('"')
|
94 |
+
review_revision += f"<div style='background-color: {highlight_color}; display: inline;'>{revision}</div>"
|
95 |
+
else:
|
96 |
+
review_revision += " " + sentence
|
97 |
+
# end revision/prepare return json
|
98 |
+
|
99 |
+
result["success"] = True
|
100 |
+
result["message"] = "Review successfully revised!"
|
101 |
+
result["data"]["revision"] = review_revision
|
102 |
+
result["data"]["score"] = review_score
|
103 |
+
result["data"]["sentence_count"] = sum(1 for sentence in sentences if len(sentence) > 20)
|
104 |
+
result["data"]["revised_sentences"] = revision_count
|
105 |
+
except Exception as e:
|
106 |
+
result["message"] = str(e)
|
107 |
+
|
108 |
+
return result
|