feat: integrate with Groq provider

chore: update deps

pyproject.toml

@@ -5,7 +5,7 @@ readme = "README.md"
 license = { text = "MIT" }
 dependencies = [
     "fastapi~=0.110.0",
-    "uvicorn[standard]~=0.28.0",
+    "uvicorn[standard]~=0.29.0",
     "promplate[all]~=0.3.3.4",
     "promplate-trace[langfuse,langsmith]==0.3.0dev2",
     "python-box~=7.1.1",
@@ -17,7 +17,7 @@ dependencies = [
     "beautifulsoup4~=4.12.3",
     "rich~=13.7.1",
     "zhipuai~=2.0.1",
-    "anthropic~=0.20.0",
+    "anthropic~=0.21.3",
     "dashscope~=1.15.0",
 ]
 

src/routes/run.py

@@ -36,7 +36,8 @@ class Msg(BaseModel):
     "claude-3-opus-20240229",
     "claude-3-sonnet-20240229",
     "claude-3-haiku-20240307",
-    "mixtral-8x7b-instruct-fp16",
+    "gemma-7b-it",
+    "mixtral-8x7b-32768",
     "nous-hermes-2-mixtral-8x7b-dpo",
     "qwen-turbo",
     "qwen-plus",

src/utils/config.py

@@ -11,6 +11,7 @@ class Config(BaseSettings):
     openai_base_url: str = ""
     octoai_api_key: str = ""
     zhipu_api_key: str = ""
+    groq_api_key: str = ""
 
     # other services
     serper_api_key: str = ""

src/utils/llm/__init__.py

@@ -5,3 +5,4 @@
 from .octoai import octoai
 from .openai import openai
 from .qwen import qwen
+from .groq import groq

src/utils/llm/groq.py

@@ -0,0 +1,44 @@
+from promplate import Message
+from promplate.llm.base import AsyncComplete, AsyncGenerate
+from promplate.llm.openai import AsyncChatComplete, AsyncChatGenerate, AsyncChatOpenAI
+from promplate_trace.auto import patch
+
+from ..config import env
+from .common import client
+from .dispatch import link_llm
+
+GROQ_BASE_URL = "https://api.groq.com/openai/v1"
+
+complete: AsyncComplete = AsyncChatComplete(http_client=client, base_url=GROQ_BASE_URL, api_key=env.groq_api_key)
+generate: AsyncGenerate = AsyncChatGenerate(http_client=client, base_url=GROQ_BASE_URL, api_key=env.groq_api_key)
+
+
+@link_llm("gemma")
+@link_llm("mixtral")
+class Groq(AsyncChatOpenAI):
+    async def complete(self, prompt: str | list[Message], /, **config):
+        config = self._run_config | config
+        return (await complete(prompt, **config)).removeprefix(" ")
+
+    async def generate(self, prompt: str | list[Message], /, **config):
+        config = self._run_config | config
+
+        first_token = True
+
+        async for token in generate(prompt, **config):
+            if token and first_token:
+                first_token = False
+                yield token.removeprefix(" ")
+            else:
+                yield token
+
+    def bind(self, **run_config):  # type: ignore
+        self._run_config.update(run_config)  # inplace
+        return self
+
+
+groq = Groq().bind(model="mixtral-8x7b-32768")
+
+
+groq.complete = patch.chat.acomplete(groq.complete)  # type: ignore
+groq.generate = patch.chat.agenerate(groq.generate)  # type: ignore

src/utils/llm/octoai.py

@@ -13,7 +13,6 @@
 generate: AsyncGenerate = AsyncChatGenerate(http_client=client, base_url=OCTOAI_BASE_URL, api_key=env.octoai_api_key)
 
 
-@link_llm("mixtral")
 @link_llm("nous-hermes")
 class OctoAI(AsyncChatOpenAI):
     async def complete(self, prompt: str | list[Message], /, **config):