Spaces:
Build error
Build error
Update app.py
Browse files
app.py
CHANGED
|
@@ -146,33 +146,18 @@ def generate_blog_post(query, prompt_template):
|
|
| 146 |
frequency_penalty = 0.5
|
| 147 |
presence_penalty = 0.3
|
| 148 |
|
| 149 |
-
# ๋ถํ์ํ ํํ
|
| 150 |
-
unwanted_phrases = [
|
| 151 |
-
'์ฌ๋ฌ๋ถ',
|
| 152 |
-
'์ต๊ทผ',
|
| 153 |
-
'๋ง์ง๋ง์ผ๋ก',
|
| 154 |
-
'๊ฒฐ๋ก ์ ์ผ๋ก',
|
| 155 |
-
'๊ฒฐ๊ตญ',
|
| 156 |
-
'์ข
ํฉ์ ์ผ๋ก',
|
| 157 |
-
'๋ฐ๋ผ์',
|
| 158 |
-
'๋ง๋ฌด๋ฆฌ',
|
| 159 |
-
'์์ฝ'
|
| 160 |
-
]
|
| 161 |
-
|
| 162 |
-
# ๋ถํ์ํ ํํ์ด ํฌํจ๋ ๋จ์ด ์ ์ฒด๋ฅผ ์ ๊ฑฐํ๋ ํจ์
|
| 163 |
def remove_unwanted_phrases(text):
|
| 164 |
-
|
|
|
|
|
|
|
|
|
|
| 165 |
words = re.findall(r'\S+|\n', text)
|
| 166 |
-
result_words = []
|
| 167 |
-
for word in words:
|
| 168 |
-
# ๋ถํ์ํ ํํ์ด ๋จ์ด์ ํฌํจ๋์ด ์๋์ง ํ์ธ
|
| 169 |
-
if not any(phrase in word for phrase in unwanted_phrases):
|
| 170 |
-
result_words.append(word)
|
| 171 |
-
# ๋จ์ด๋ค์ ๋ค์ ์กฐํฉํ์ฌ ๊ฒฐ๊ณผ ํ
์คํธ ์์ฑ
|
| 172 |
return ' '.join(result_words).replace(' \n ', '\n').replace(' \n', '\n').replace('\n ', '\n')
|
| 173 |
|
| 174 |
# ์ด๊ธฐ ํ๋กฌํํธ ๊ตฌ์ฑ
|
| 175 |
-
|
| 176 |
์ฃผ์ : {query}
|
| 177 |
์ฐธ๊ณ ๊ธ 1: {ref1}
|
| 178 |
์ฐธ๊ณ ๊ธ 2: {ref2}
|
|
@@ -180,56 +165,55 @@ def generate_blog_post(query, prompt_template):
|
|
| 180 |
๋ชฉํ ๊ธ์์: {target_char_length}
|
| 181 |
"""
|
| 182 |
|
| 183 |
-
|
| 184 |
-
|
| 185 |
-
|
| 186 |
-
|
| 187 |
-
|
| 188 |
-
|
| 189 |
-
|
| 190 |
-
|
| 191 |
-
|
| 192 |
-
|
| 193 |
-
|
| 194 |
-
|
| 195 |
-
|
| 196 |
-
|
| 197 |
-
|
| 198 |
-
|
| 199 |
-
|
| 200 |
-
|
| 201 |
-
|
| 202 |
-
|
| 203 |
-
|
| 204 |
-
|
| 205 |
-
|
| 206 |
-
|
| 207 |
-
max_tokens=max_tokens,
|
| 208 |
-
top_p=top_p,
|
| 209 |
-
frequency_penalty=frequency_penalty,
|
| 210 |
-
presence_penalty=presence_penalty,
|
| 211 |
-
)
|
| 212 |
-
|
| 213 |
-
# ๋ชจ๋ธ์ด ์์ฑํ ๋ต๋ณ ์ถ์ถ
|
| 214 |
-
generated_post = response['choices'][0]['message']['content'].strip()
|
| 215 |
|
| 216 |
-
|
| 217 |
-
|
| 218 |
-
|
| 219 |
-
|
| 220 |
-
actual_char_length = len(generated_post)
|
| 221 |
|
| 222 |
-
|
| 223 |
-
|
| 224 |
-
|
| 225 |
-
|
| 226 |
-
|
| 227 |
-
|
| 228 |
-
|
| 229 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 230 |
|
| 231 |
# ์ต์ข
๊ฒฐ๊ณผ๋ฌผ ๊ตฌ์ฑ
|
| 232 |
-
final_post = f"์ฃผ์ : {query}\n\n{
|
|
|
|
| 233 |
|
| 234 |
return final_post, ref1, ref2, ref3, actual_char_length
|
| 235 |
|
|
|
|
| 146 |
frequency_penalty = 0.5
|
| 147 |
presence_penalty = 0.3
|
| 148 |
|
| 149 |
+
# ๋ถํ์ํ ํํ ์ ๊ฑฐ ํจ์
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 150 |
def remove_unwanted_phrases(text):
|
| 151 |
+
unwanted_phrases = [
|
| 152 |
+
'์ฌ๋ฌ๋ถ', '์ต๊ทผ', '๋ง์ง๋ง์ผ๋ก', '๊ฒฐ๋ก ์ ์ผ๋ก', '๊ฒฐ๊ตญ',
|
| 153 |
+
'์ข
ํฉ์ ์ผ๋ก', '๋ฐ๋ผ์', '๋ง๋ฌด๋ฆฌ', '์์ฝ'
|
| 154 |
+
]
|
| 155 |
words = re.findall(r'\S+|\n', text)
|
| 156 |
+
result_words = [word for word in words if not any(phrase in word for phrase in unwanted_phrases)]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 157 |
return ' '.join(result_words).replace(' \n ', '\n').replace(' \n', '\n').replace('\n ', '\n')
|
| 158 |
|
| 159 |
# ์ด๊ธฐ ํ๋กฌํํธ ๊ตฌ์ฑ
|
| 160 |
+
initial_prompt = prompt_template + f"""
|
| 161 |
์ฃผ์ : {query}
|
| 162 |
์ฐธ๊ณ ๊ธ 1: {ref1}
|
| 163 |
์ฐธ๊ณ ๊ธ 2: {ref2}
|
|
|
|
| 165 |
๋ชฉํ ๊ธ์์: {target_char_length}
|
| 166 |
"""
|
| 167 |
|
| 168 |
+
# ์ฒซ ๋ฒ์งธ ์๋
|
| 169 |
+
messages = [{"role": "user", "content": initial_prompt}]
|
| 170 |
+
response = openai.ChatCompletion.create(
|
| 171 |
+
model=model_name,
|
| 172 |
+
messages=messages,
|
| 173 |
+
temperature=temperature,
|
| 174 |
+
max_tokens=max_tokens,
|
| 175 |
+
top_p=top_p,
|
| 176 |
+
frequency_penalty=frequency_penalty,
|
| 177 |
+
presence_penalty=presence_penalty,
|
| 178 |
+
)
|
| 179 |
+
first_attempt = response['choices'][0]['message']['content'].strip()
|
| 180 |
+
|
| 181 |
+
# ๋ ๋ฒ์งธ ์๋๋ฅผ ์ํ ์ถ๊ฐ ํ๋กฌํํธ
|
| 182 |
+
additional_prompt = f"""
|
| 183 |
+
์ด์ ์ ์์ฑ๋ ๊ธ์ ๊ธฐ๋ฐ์ผ๋ก ๋ค์ ์ง์นจ์ ๋ฐ๋์ ๋ฐ๋ผ์ ๊ธ์ ํด๊ณ (revision)ํ๋ผ
|
| 184 |
+
1. ๋ฐ๋์ ์ด์ ๊ธ(first_attempt)์ ๊ตฌ์กฐ์ ๋ด์ฉ์ ์ ์งํ๋ฉด์ ๋ด์ฉ์ ๋ณด์ํ๋ผ
|
| 185 |
+
2. ๋ฐ๋์ ๋งํฌ๋ค์ด ํ์์ด ์๋ ์์ํ ํ
์คํธ๋ก๋ง ์ถ๋ ฅํ์ธ์.
|
| 186 |
+
3. ๋ฐ๋์ ์ด ํํ๋ค์ ์ฌ์ฉํ์ง ๋ง์ธ์: ์ฌ๋ฌ๋ถ, ์ต๊ทผ, ๋ง์ง๋ง์ผ๋ก, ๊ฒฐ๋ก ์ ์ผ๋ก, ๊ฒฐ๊ตญ, ์ข
ํฉ์ ์ผ๋ก, ๋ฐ๋ผ์, ๋ง๋ฌด๋ฆฌ, ์์ฝ.
|
| 187 |
+
4. ๊ธ์ ํ๋ฆ์ ์์ฐ์ค๋ฝ๊ฒ ๋ง๋ค๊ณ , ๊ฐ ๋จ๋ฝ ๊ฐ์ ์ฐ๊ฒฐ์ ๋ถ๋๋ฝ๊ฒ ํด์ฃผ์ธ์.
|
| 188 |
+
5. ๋ฐ๋์ ๋ชฉํ ๊ธ์์({target_char_length}์)๋ฅผ ์ถฉ์กฑํ๋๋ก ๋ด์ฉ์ ํ์ฅํ์ธ์.
|
| 189 |
+
|
| 190 |
+
์ด์ ๊ธ:
|
| 191 |
+
{first_attempt}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 192 |
|
| 193 |
+
์ฐธ๊ณ ๊ธ 1: {ref1}
|
| 194 |
+
์ฐธ๊ณ ๊ธ 2: {ref2}
|
| 195 |
+
์ฐธ๊ณ ๊ธ 3: {ref3}
|
| 196 |
+
"""
|
|
|
|
| 197 |
|
| 198 |
+
# ๋ ๋ฒ์งธ ์๋
|
| 199 |
+
messages = [{"role": "user", "content": additional_prompt}]
|
| 200 |
+
response = openai.ChatCompletion.create(
|
| 201 |
+
model=model_name,
|
| 202 |
+
messages=messages,
|
| 203 |
+
temperature=temperature,
|
| 204 |
+
max_tokens=max_tokens,
|
| 205 |
+
top_p=top_p,
|
| 206 |
+
frequency_penalty=frequency_penalty,
|
| 207 |
+
presence_penalty=presence_penalty,
|
| 208 |
+
)
|
| 209 |
+
second_attempt = response['choices'][0]['message']['content'].strip()
|
| 210 |
+
|
| 211 |
+
# ๋ถํ์ํ ํํ ์ ๊ฑฐ
|
| 212 |
+
final_post = remove_unwanted_phrases(second_attempt)
|
| 213 |
|
| 214 |
# ์ต์ข
๊ฒฐ๊ณผ๋ฌผ ๊ตฌ์ฑ
|
| 215 |
+
final_post = f"์ฃผ์ : {query}\n\n{final_post}"
|
| 216 |
+
actual_char_length = len(final_post)
|
| 217 |
|
| 218 |
return final_post, ref1, ref2, ref3, actual_char_length
|
| 219 |
|