Spaces:
Build error
Build error
update gpt3.5
Browse files
app.py
CHANGED
|
@@ -61,10 +61,29 @@ def call_openai_summary(openaiobj,transcription):
|
|
| 61 |
]
|
| 62 |
)
|
| 63 |
return response.choices[0].message.content
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 64 |
|
| 65 |
|
| 66 |
|
| 67 |
-
def split_into_chunks(text, tokens=
|
| 68 |
#encoding = tiktoken.encoding_for_model('gpt-3.5-turbo')
|
| 69 |
encoding = tiktoken.encoding_for_model(usemodelname)
|
| 70 |
words = encoding.encode(text)
|
|
@@ -92,15 +111,18 @@ def process_chunks(openaikeystr,inputtext):
|
|
| 92 |
chunks = split_into_chunks(text)
|
| 93 |
response='這是分段會議紀錄結果\n\n'
|
| 94 |
i=1
|
| 95 |
-
|
|
|
|
|
|
|
| 96 |
|
| 97 |
-
|
| 98 |
-
|
| 99 |
-
# response=response+call_openai_summary(openaiobj,chunk)
|
| 100 |
-
if i>2:
|
| 101 |
finalresponse=response+'\n\n 這是根據以上分段會議紀錄彙編如下 \n\n' +call_openai_api(openaiobj,response)
|
|
|
|
|
|
|
|
|
|
| 102 |
else:
|
| 103 |
-
finalresponse=
|
| 104 |
return finalresponse
|
| 105 |
# # Processes chunks in parallel
|
| 106 |
# with ThreadPoolExecutor() as executor:
|
|
@@ -212,7 +234,7 @@ file_transcribe = gr.Interface(
|
|
| 212 |
allow_flagging="never",
|
| 213 |
)
|
| 214 |
import google.generativeai as genai
|
| 215 |
-
def gpt4write(
|
| 216 |
# openaiobj = OpenAI(
|
| 217 |
# # This is the default and can be omitted
|
| 218 |
|
|
|
|
| 61 |
]
|
| 62 |
)
|
| 63 |
return response.choices[0].message.content
|
| 64 |
+
def call_openai_summaryall(openaiobj,transcription):
|
| 65 |
+
|
| 66 |
+
response = openaiobj.chat.completions.create(
|
| 67 |
+
#model="gpt-3.5-turbo",
|
| 68 |
+
model=usemodelname,
|
| 69 |
+
temperature=0,
|
| 70 |
+
messages=[
|
| 71 |
+
{
|
| 72 |
+
"role": "system",
|
| 73 |
+
"content": "你是專業的會議紀錄製作員,請根據分段的會議決證,彙整成正式會議紀錄"
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"role": "user",
|
| 77 |
+
"content": transcription
|
| 78 |
+
}
|
| 79 |
+
]
|
| 80 |
+
)
|
| 81 |
+
return response.choices[0].message.content
|
| 82 |
+
|
| 83 |
|
| 84 |
|
| 85 |
|
| 86 |
+
def split_into_chunks(text, tokens=15900):
|
| 87 |
#encoding = tiktoken.encoding_for_model('gpt-3.5-turbo')
|
| 88 |
encoding = tiktoken.encoding_for_model(usemodelname)
|
| 89 |
words = encoding.encode(text)
|
|
|
|
| 111 |
chunks = split_into_chunks(text)
|
| 112 |
response='這是分段會議紀錄結果\n\n'
|
| 113 |
i=1
|
| 114 |
+
if len(chunks)>1:
|
| 115 |
+
|
| 116 |
+
for chunk in chunks:
|
| 117 |
|
| 118 |
+
response=response+'第' +str(i)+'段\n'+call_openai_api(openaiobj,chunk)+'\n\n'
|
| 119 |
+
i=i+1
|
|
|
|
|
|
|
| 120 |
finalresponse=response+'\n\n 這是根據以上分段會議紀錄彙編如下 \n\n' +call_openai_api(openaiobj,response)
|
| 121 |
+
# response=response+call_openai_summary(openaiobj,chunk)
|
| 122 |
+
|
| 123 |
+
|
| 124 |
else:
|
| 125 |
+
finalresponse=call_openai_api(openaiobj,chunk[0])
|
| 126 |
return finalresponse
|
| 127 |
# # Processes chunks in parallel
|
| 128 |
# with ThreadPoolExecutor() as executor:
|
|
|
|
| 234 |
allow_flagging="never",
|
| 235 |
)
|
| 236 |
import google.generativeai as genai
|
| 237 |
+
def gpt4write(openaikeystr,transcribe_text):
|
| 238 |
# openaiobj = OpenAI(
|
| 239 |
# # This is the default and can be omitted
|
| 240 |
|