1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
|
from __future__ import annotations
from aiohttp import ClientSession
import json
from ...typing import AsyncResult, Messages
from ..base_provider import AsyncGeneratorProvider, ProviderModelMixin
from ..helper import format_prompt
class NexraLLaMA31(AsyncGeneratorProvider, ProviderModelMixin):
label = "Nexra LLaMA 3.1"
url = "https://nexra.aryahcr.cc/documentation/llama-3.1/en"
api_endpoint = "https://nexra.aryahcr.cc/api/chat/complements"
working = True
supports_stream = True
default_model = 'llama-3.1'
models = [default_model]
@classmethod
def get_model(cls, model: str) -> str:
return cls.default_model
@classmethod
async def create_async_generator(
cls,
model: str,
messages: Messages,
proxy: str = None,
stream: bool = False,
markdown: bool = False,
**kwargs
) -> AsyncResult:
model = cls.get_model(model)
headers = {
"Content-Type": "application/json"
}
async with ClientSession(headers=headers) as session:
prompt = format_prompt(messages)
data = {
"messages": [
{
"role": "user",
"content": prompt
}
],
"stream": stream,
"markdown": markdown,
"model": model
}
async with session.post(f"{cls.api_endpoint}", json=data, proxy=proxy) as response:
response.raise_for_status()
if stream:
# Streamed response handling
collected_message = ""
async for chunk in response.content.iter_any():
if chunk:
decoded_chunk = chunk.decode().strip().split("\x1e")
for part in decoded_chunk:
if part:
message_data = json.loads(part)
# Collect messages until 'finish': true
if 'message' in message_data and message_data['message']:
collected_message = message_data['message']
# When finish is true, yield the final collected message
if message_data.get('finish', False):
yield collected_message
return
else:
# Non-streamed response handling
response_data = await response.json(content_type=None)
# Yield the message directly from the response
if 'message' in response_data and response_data['message']:
yield response_data['message']
return
|
