vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

oai proxy: workaround for some client (such as Chatbox)

Browse source
This commit is contained in:
rhjdvsgsgks 2023-11-08 12:20:40 +00:00 • committed by GitHub
parent 0f3e6b0bba
commit 73bb25e901
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 3 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

6
examples/server/api_like_OAI.py
View file

@ -179,12 +179,12 @@ def chat_completions():
data = requests.request("POST", urllib.parse.urljoin(args.llama_api, "/completion"), data=json.dumps(postData), stream=True) data = requests.request("POST", urllib.parse.urljoin(args.llama_api, "/completion"), data=json.dumps(postData), stream=True)
time_now = int(time.time()) time_now = int(time.time())
resData = make_resData_stream({}, chat=True, time_now=time_now, start=True) resData = make_resData_stream({}, chat=True, time_now=time_now, start=True)
yield 'data: {}\n'.format(json.dumps(resData)) yield 'data: {}\n\n'.format(json.dumps(resData))
for line in data.iter_lines(): for line in data.iter_lines():
if line: if line:
decoded_line = line.decode('utf-8') decoded_line = line.decode('utf-8')
resData = make_resData_stream(json.loads(decoded_line[6:]), chat=True, time_now=time_now) resData = make_resData_stream(json.loads(decoded_line[6:]), chat=True, time_now=time_now)
yield 'data: {}\n'.format(json.dumps(resData)) yield 'data: {}\n\n'.format(json.dumps(resData))
return Response(generate(), mimetype='text/event-stream', headers={"Access-Control-Allow-Origin": "*", "Access-Control-Allow-Headers": "*"}) return Response(generate(), mimetype='text/event-stream', headers={"Access-Control-Allow-Origin": "*", "Access-Control-Allow-Headers": "*"})
@ -220,7 +220,7 @@ def completion():
if line: if line:
decoded_line = line.decode('utf-8') decoded_line = line.decode('utf-8')
resData = make_resData_stream(json.loads(decoded_line[6:]), chat=False, time_now=time_now) resData = make_resData_stream(json.loads(decoded_line[6:]), chat=False, time_now=time_now)
yield 'data: {}\n'.format(json.dumps(resData)) yield 'data: {}\n\n'.format(json.dumps(resData))
return Response(generate(), mimetype='text/event-stream', headers={"Access-Control-Allow-Origin": "*", "Access-Control-Allow-Headers": "*"}) return Response(generate(), mimetype='text/event-stream', headers={"Access-Control-Allow-Origin": "*", "Access-Control-Allow-Headers": "*"})
if __name__ == '__main__': if __name__ == '__main__':