Spaces:
Sleeping
Sleeping
File size: 8,658 Bytes
c1fe6d3 |
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 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 |
import typing as t
import asyncio
from abc import ABC, abstractmethod
import os
# External Libraries
import requests
import json
from together import Together
from openai import AsyncOpenAI
import aiohttp # Using aiohttp for async HTTP requests
# A standard response type to avoid type errors with Together.
try:
from together.types.chat_completions import ChatCompletionResponse
except ImportError:
ChatCompletionResponse = t.Any
# --- ABSTRACT BASE CLASS (The Core Abstraction) ---
class BaseLLMClient(ABC):
"""
Abstract base class for all LLM clients.
Defines a common, asynchronous interface for calling LLMs.
"""
def __init__(self, model: str, **kwargs):
self.client = None
self.model = model
self.kwargs = kwargs
@abstractmethod
async def __call__(
self,
prompt: str,
max_tokens: int = 1_000,
system_prompt: t.Optional[str] = None,
**kwargs
) -> str:
"""
Defines the async call method for all concrete clients.
"""
...
def _create_messages(
self,
prompt: str,
system_prompt: t.Optional[str] = None
) -> t.List[t.Dict[str, str]]:
"""
Helper to create a standard message dictionary with an optional system prompt.
The system prompt is always added as the first message to set the model's context.
"""
messages = []
if system_prompt:
messages.append({"role": "system", "content": system_prompt})
messages.append({"role": "user", "content": prompt})
return messages
# --- CONCRETE IMPLEMENTATIONS (The Adapters) ---
class OpenAIClient(BaseLLMClient):
"""
Adapter for the OpenAI (and OpenAI-compatible) Async API client.
"""
def __init__(self, model: str, **kwargs):
super().__init__(model, **kwargs)
self.client = AsyncOpenAI(api_key=os.getenv("OPENAI_API_KEY"))
async def __call__(
self,
prompt: str,
max_tokens: int = 1_000,
system_prompt: t.Optional[str] = None,
**kwargs
) -> str:
try:
messages = self._create_messages(prompt, system_prompt)
response = await self.client.chat.completions.create(
model=self.model,
messages=messages,
max_tokens=max_tokens,
**self.kwargs,
**kwargs,
)
return response.choices[0].message.content
except Exception as e:
return f"Error from OpenAI: {e}"
class TogetherAIClient(BaseLLMClient):
"""
Adapter for the Together API client.
Uses asyncio.to_thread to run the synchronous client in a separate thread.
"""
def __init__(self, model: str, **kwargs):
super().__init__(model, **kwargs)
# Note: Together() automatically looks for TOGETHER_API_KEY env var
self.client = Together()
async def __call__(
self,
prompt: str,
max_tokens: int = 1_000,
system_prompt: t.Optional[str] = None,
**kwargs
) -> str:
# Use asyncio.to_thread to run the synchronous Together client
# without blocking the event loop.
try:
messages = self._create_messages(prompt, system_prompt)
response: ChatCompletionResponse = await asyncio.to_thread(
self.client.chat.completions.create,
model=self.model,
messages=messages,
max_tokens=max_tokens,
**self.kwargs,
**kwargs,
)
return str(response.choices[0].message.content)
except Exception as e:
return f"Error from TogetherAI: {e}"
class GeminiClient(BaseLLMClient):
"""
Adapter for the Gemini REST API, using aiohttp for async HTTP requests.
"""
def __init__(self, model: str, **kwargs):
super().__init__(model, **kwargs)
self.api_key = os.getenv("GEMINI_API_KEY")
self.url = f"https://generativelanguage.googleapis.com/v1beta/models/{self.model}:generateContent?key={self.api_key}"
async def __call__(
self,
prompt: str,
max_tokens: int = 1_000,
system_prompt: t.Optional[str] = None,
**kwargs
) -> str:
if not self.api_key:
return "Error: GEMINI_API_KEY not found."
contents = self._create_messages(prompt, system_prompt)
payload = {
"contents": contents,
"generationConfig": {"maxOutputTokens": max_tokens},
**self.kwargs,
**kwargs,
}
headers = {"Content-Type": "application/json"}
try:
async with aiohttp.ClientSession() as session:
async with session.post(self.url, headers=headers, json=payload) as response:
response.raise_for_status()
response_data = await response.json()
return response_data['candidates'][0]['content']['parts'][0]['text']
except aiohttp.ClientError as e:
return f"Error from Gemini (requests): {e}"
except (KeyError, IndexError) as e:
return f"Error parsing Gemini response: {e}"
class GroqClient(BaseLLMClient):
"""
Adapter for the Groq REST API, using aiohttp for async HTTP requests.
"""
def __init__(self, model: str, **kwargs):
super().__init__(model, **kwargs)
self.api_key = os.getenv("GROQ_API_KEY")
self.url = "https://api.groq.com/openai/v1/chat/completions"
async def __call__(
self,
prompt: str,
max_tokens: int = 1_000,
system_prompt: t.Optional[str] = None,
**kwargs
) -> str:
if not self.api_key:
return "Error: GROQ_API_KEY not found."
messages = self._create_messages(prompt, system_prompt)
payload = {
"model": self.model,
"messages": messages,
"max_tokens": max_tokens,
**self.kwargs,
**kwargs
}
headers = {
"Authorization": f"Bearer {self.api_key}",
"Content-Type": "application/json"
}
try:
async with aiohttp.ClientSession() as session:
async with session.post(self.url, headers=headers, json=payload) as response:
response.raise_for_status()
response_data = await response.json()
return response_data['choices'][0]['message']['content']
except aiohttp.ClientError as e:
return f"Error from Groq (requests): {e}"
except (KeyError, IndexError) as e:
return f"Error parsing Groq response: {e}"
class MistralClient(BaseLLMClient):
"""
Adapter for the Mistral REST API, using aiohttp for async HTTP requests.
"""
def __init__(self, model: str, **kwargs):
super().__init__(model, **kwargs)
self.api_key = os.getenv("MISTRAL_API_KEY")
self.url = "https://api.mistral.ai/v1/chat/completions"
async def __call__(
self,
prompt: str,
max_tokens: int = 1_000,
system_prompt: t.Optional[str] = None,
**kwargs
) -> str:
if not self.api_key:
return "Error: MISTRAL_API_KEY not found."
messages = self._create_messages(prompt, system_prompt)
payload = {
"model": self.model,
"messages": messages,
"max_tokens": max_tokens,
**self.kwargs,
**kwargs
}
headers = {
"Authorization": f"Bearer {self.api_key}",
"Content-Type": "application/json"
}
try:
async with aiohttp.ClientSession() as session:
async with session.post(self.url, headers=headers, json=payload) as response:
response.raise_for_status()
response_data = await response.json()
return response_data['choices'][0]['message']['content']
except aiohttp.ClientError as e:
return f"Error from Mistral (requests): {e}"
except (KeyError, IndexError) as e:
return f"Error parsing Mistral response: {e}"
# ('openai', OpenAIClient(model="gpt-3.5-turbo")),
# ('togetherai', TogetherAIClient(model="meta-llama/Llama-3.3-70B-Instruct-Turbo-Free")),
# ('gemini', GeminiClient(model="gemini-1.5-flash-latest")),
# ('groq', GroqClient(model="llama3-8b-8192")),
# ('mistral', MistralClient(model="mistral-tiny")), |