Merge branch 'main' into deploy

pyproject.toml

@@ -4,9 +4,9 @@ requires-python = ">=3.10,<3.13"
 readme = "README.md"
 license = { text = "MIT" }
 dependencies = [
-    "fastapi~=0.110.1",
+    "fastapi~=0.110.3",
     "uvicorn[standard]~=0.29.0",
-    "promplate[all]~=0.3.4.2",
+    "promplate[all]~=0.3.4.4",
     "promplate-trace[langfuse,langsmith]==0.3.0dev2",
     "python-box~=7.1.1",
     "pydantic-settings~=2.2.1",
@@ -17,14 +17,14 @@ dependencies = [
     "beautifulsoup4~=4.12.3",
     "rich~=13.7.1",
     "zhipuai~=2.0.1",
-    "anthropic~=0.25.2",
-    "dashscope~=1.17.0",
+    "anthropic~=0.25.7",
+    "dashscope~=1.17.1",
 ]
 
 [tool.pdm.dev-dependencies]
 dev = [
     "isort~=5.13.2",
-    "black~=24.4.0",
+    "black~=24.4.2",
 ]
 
 [tool.pdm.scripts]

src/routes/run.py

@@ -12,7 +12,7 @@
 from ..logic.tools import tool_map
 from ..utils.config import env
 from ..utils.http import forward_headers
-from ..utils.llm import find_llm
+from ..utils.llm import find_llm, groq, openai
 from .sse import non_duplicated_event_stream
 
 run_router = APIRouter(tags=["call"])
@@ -110,7 +110,7 @@ async def make_stream():
 
         config = data.config
 
-        if data.model.startswith("gpt"):
+        if find_llm(data.model) in (openai, groq):
             config["extra_headers"] = forward_headers(r.headers)
 
         async for c in node.astream(data.context, find_llm(data.model).generate, **config):

src/utils/config.py

@@ -12,6 +12,7 @@ class Config(BaseSettings):
     octoai_api_key: str = ""
     zhipu_api_key: str = ""
     groq_api_key: str = ""
+    groq_base_url: str = "https://api.groq.com/openai/v1"
 
     # other services
     serper_api_key: str = ""

src/utils/llm/groq.py

@@ -7,10 +7,8 @@
 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)
+complete: AsyncComplete = AsyncChatComplete(http_client=client, base_url=env.groq_base_url, api_key=env.groq_api_key)
+generate: AsyncGenerate = AsyncChatGenerate(http_client=client, base_url=env.groq_base_url, api_key=env.groq_api_key)
 
 
 @link_llm("gemma")