Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[ref] optimize the API calling of InternLM2-chat-7b #54

Merged
merged 1 commit into from
Jan 26, 2024
Merged

[ref] optimize the API calling of InternLM2-chat-7b #54

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions openaoe/backend/model/internlm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,11 +20,11 @@ class InternlmChatCompletionBody(BaseModel):
prompt: Optional[str]
messages: List[Context]
temperature: Optional[float] = 0.7
top_p: Optional[float] = 1
top_p: Optional[float] = 0.8
n: Optional[int] = 1
stop: Optional[Union[str, List]] = "false"
presence_penalty: Optional[float] = 0
frequency_penalty: Optional[float] = 0
frequency_penalty: Optional[float] = 0.5
user: Optional[str] = NOT_GIVEN
stream: Optional[bool] = False
timeout: Optional[int] = 600
Expand Down
70 changes: 41 additions & 29 deletions openaoe/backend/service/service_internlm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,7 +42,6 @@ def chat_completion_v1(request, body: InternlmChatCompletionBody):
"top_p": body.top_p,
"n": body.n,
"max_tokens": body.max_tokens,
"stop": "false",
"stream": body.stream,
"presence_penalty": body.presence_penalty,
"frequency_penalty": body.frequency_penalty,
Expand Down Expand Up @@ -73,40 +72,53 @@ async def event_generator_json(count: int):
while count > 0:
count -= 1
stop_flag = False
response = ""
if await request.is_disconnected():
break
try:
res = requests.post(url, headers=headers, json=data)
res_data = res.text.replace("data: ", "")
json_strings = res_data.split("\n\n")
for chunk in json_strings:
try:
json_data = json.loads(chunk)
choice = json_data["choices"][0]
if choice['finish_reason']:
stop_flag = True
if 'content' not in choice['delta']:
continue
s = choice["delta"]["content"]
if s:
response += s
dict_item = {
"success": "true",
"msg": s
}
yield json.dumps(dict_item, ensure_ascii=False)
if stop_flag:
break
except json.JSONDecodeError as e:
print(f"JSON Parse Error: {e}")

if stop_flag:
break
res = requests.post(url, headers=headers, json=data, stream=True)
logger.debug(f"url={url}, headers={headers}, body={data}")
if res:
for chunk in res.iter_content(chunk_size=512, decode_unicode=True):
logger.debug(f"chunk: {chunk}")
res_data = chunk.replace("data: ", "")
try:
json_data = json.loads(res_data)
logger.debug(f"json_data: {json_data}")
if json_data.get("object") == "error":
error_msg = json_data.get("message")
dict_item = {
"success": "true",
"msg": f"LMDeploy: {error_msg} "
}
yield json.dumps(dict_item, ensure_ascii=False)
stop_flag = True
choices = json_data.get("choices")
if not choices:
continue
choice = choices[0]
if choice.get('finish_reason'):
stop_flag = True
if 'content' not in choice['delta']:
continue
s = choice["delta"]["content"]
logger.debug(f"content={s}")
if s:
dict_item = {
"success": "true",
"msg": s
}
yield json.dumps(dict_item, ensure_ascii=False)
if stop_flag:
break
except Exception as e:
logger.error(f"Error: {e}")
if stop_flag:
break
except Exception as e:
logger.error(f"{e}")
yield json.dumps({
"success": "false",
"msg": str(e)
"msg": f"from backend: {str(e)}"
})
break

Expand Down