alex/nova-api
1
0
Fork 0
mirror of https://github.com/NovaOSS/nova-api.git synced 2024-11-25 18:53:58 +01:00
Code Issues Packages Projects Releases Wiki Activity

Compare commits

base: alex:fffd7a276a455466e65e84b81f8415af244a1fde
Branches Tags
alex:main
alex:gpt-3.5-turbo-1106
alex:PullGPT/issue-8
alex:tokenbasedcost
..
compare: alex:a16ba559fbfae5a125fe224208f8d119211d6523
Branches Tags
alex:gpt-3.5-turbo-1106
alex:main
alex:PullGPT/issue-8
alex:tokenbasedcost

No commits in common. "fffd7a276a455466e65e84b81f8415af244a1fde" and "a16ba559fbfae5a125fe224208f8d119211d6523" have entirely different histories.
fffd7a276a ... a16ba559fb

3 changed files with 9 additions and 13 deletions
Show stats Expand all files Collapse all files

4
api/providers/__init__.py
View file

@ -1,2 +1,2 @@
from . import ails, closed, closed4 from . import ails
MODULES = [closed, closed4] MODULES = [ails]

16
api/responder.py
View file

@ -68,8 +68,7 @@ async def respond(
'insufficient_quota': 0, 'insufficient_quota': 0,
'billing_not_active': 0, 'billing_not_active': 0,
'critical_provider_error': 0, 'critical_provider_error': 0,
'timeout': 0, 'timeout': 0
'other_errors': []
} }
input_tokens = 0 input_tokens = 0
@ -169,11 +168,6 @@ async def respond(
output_tokens = client_json_response['usage']['completion_tokens'] output_tokens = client_json_response['usage']['completion_tokens']
server_json_response = client_json_response server_json_response = client_json_response
elif response.content_type == 'text/plain':
data = (await response.read()).decode("utf-8")
print(f'[!] {data}')
skipped_errors['other_errors'] = skipped_errors['other_errors'].append(data)
continue
if is_stream: if is_stream:
input_tokens = await count_tokens_for_messages(payload['messages'], model=model) input_tokens = await count_tokens_for_messages(payload['messages'], model=model)
@ -181,7 +175,7 @@ async def respond(
chunk_no = 0 chunk_no = 0
buffer = '' buffer = ''
async for chunk in response.content.iter_chunked(1024): async for chunk in response.content.iter_chunked(1024):
chunk_no += 1 chunk_no += 1
chunk = chunk.decode('utf8') chunk = chunk.decode('utf8')
@ -211,8 +205,7 @@ async def respond(
continue continue
else: else:
skipped_errors = {k: v for k, v in skipped_errors.items() if ((isinstance(v, int) and v > 0) or skipped_errors = {k: v for k, v in skipped_errors.items() if v > 0}
(isinstance(v, list) and len(v) > 0))}
skipped_errors = ujson.dumps(skipped_errors, indent=4) skipped_errors = ujson.dumps(skipped_errors, indent=4)
yield await errors.yield_error(500, yield await errors.yield_error(500,
f'Sorry, our API seems to have issues connecting to "{model}".', f'Sorry, our API seems to have issues connecting to "{model}".',
@ -223,6 +216,7 @@ async def respond(
if (not is_stream) and server_json_response: if (not is_stream) and server_json_response:
yield json.dumps(server_json_response) yield json.dumps(server_json_response)
role = user.get('role', 'default') role = user.get('role', 'default')
model_multipliers = config['costs'] model_multipliers = config['costs']
@ -257,6 +251,8 @@ async def respond(
credits_cost = round(credits_cost * role_cost_multiplier) credits_cost = round(credits_cost * role_cost_multiplier)
print(f'[bold]Credits cost[/bold]: {credits_cost}')
create_background_task( create_background_task(
after_request.after_request( after_request.after_request(
provider=provider_name, provider=provider_name,

2
checks/client.py
View file

@ -222,7 +222,7 @@ async def demo():
for func in [ for func in [
test_chat_stream_gpt3, test_chat_stream_gpt3,
test_chat_non_stream_gpt4, test_chat_non_stream_gpt4,
# test_function_calling, test_function_calling,
]: ]:
print(f'[*] {func.__name__}') print(f'[*] {func.__name__}')
result = await func() result = await func()