dev branch updates

.github/workflows/build_and_push.yml

@@ -1,7 +1,7 @@
 name: Staging Build and Deploy
 on:
   push:
-    branches: ["master"]
+    branches: ["dev"]
 
 jobs:
   build-and-push:
@@ -28,6 +28,17 @@ jobs:
         id: iname
         run: echo "iname=firstbatch/${{ github.event.repository.name }}" >> $GITHUB_OUTPUT
 
+      - name: Remove unnecessary files # https://github.com/actions/runner-images/issues/2840#issuecomment-1284059930
+        run: |
+          sudo rm -rf /usr/share/dotnet
+          sudo rm -rf "$AGENT_TOOLSDIRECTORY"
+
+      - name: Login to Docker Hub
+        uses: docker/login-action@v3
+        with:
+          username: ${{ secrets.DOCKERHUB_USERNAME }}
+          password: ${{ secrets.DOCKERHUB_TOKEN }}
+
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v2
         with:
@@ -39,14 +50,9 @@ jobs:
           IMAGE_TAG: ${{ steps.itag.outputs.itag }}
         run: |
           docker buildx create --use --name ibuilder --driver docker-container
-
-          docker buildx build --platform linux/arm64 --builder=ibuilder -t $IMAGE_NAME:$IMAGE_TAG-arm64 --load .
-          docker push $IMAGE_NAME:$IMAGE_TAG-arm64
-          docker rmi $IMAGE_NAME:$IMAGE_TAG-arm64
-
-          docker buildx build --platform linux/amd64 --builder=ibuilder -t $IMAGE_NAME:$IMAGE_TAG-amd64 --load .
-          docker push $IMAGE_NAME:$IMAGE_TAG-amd64
-          docker rmi $IMAGE_NAME:$IMAGE_TAG-amd64
+          docker buildx build --platform linux/arm64 --builder=ibuilder -t $IMAGE_NAME:$IMAGE_TAG --load .
+          docker push $IMAGE_NAME:$IMAGE_TAG
+          docker rmi $IMAGE_NAME:$IMAGE_TAG
 
     outputs:
       itag: ${{ steps.itag.outputs.itag }}

.gitignore

@@ -1,3 +1,7 @@
 .env
 /.idea/.gitignore
 /playground.py
+.DS_Store
+/dist
+/qdrant_storage
+*__pycache__*

docker-compose.yml

@@ -31,15 +31,15 @@ services:
     profiles: [ollama-docker]
 
   search-agent-server:
-    image: dria-searching-agent:server
+    image: firstbatch/dria-searching-agent:latest
     build:
       context: .
       dockerfile: Dockerfile
     ports:
       - 5000:5000
     environment:
-      AGENT_MODEL_PROVIDER: Ollama
-      AGENT_MODEL: phi3:latest
+      AGENT_MODEL_PROVIDER: ${AGENT_MODEL_PROVIDER}
+      AGENT_MODEL_NAME: ${AGENT_MODEL_NAME}
 
       ANTHROPIC_API_KEY: ${ANTHROPIC_API_KEY}
       OPENAI_API_KEY: ${OPENAI_API_KEY}

src/dria_searching_agent/config/config.py

@@ -5,22 +5,21 @@ class Config:
     def __init__(self):
         load_dotenv()  # This loads environment variables from a .env file if it exists
 
-        self.agent_model_provider = os.getenv('AGENT_MODEL_PROVIDER')
-        self.agent_model = os.getenv('AGENT_MODEL')
+        self.agent_model_provider = os.getenv('AGENT_MODEL_PROVIDER', "ollama")
+        self.agent_model_name = os.getenv('AGENT_MODEL_NAME', "gpt-4o")
+        self.agent_max_iter = os.getenv('AGENT_MAX_ITER', 10)
 
         self.anthropic_key = os.getenv('ANTHROPIC_KEY')
         self.openai_api_key = os.getenv('OPENAI_API_KEY')
         self.serper_api_key = os.getenv('SERPER_API_KEY')
         self.browserless_token = os.getenv('BROWSERLESS_TOKEN')
 
-        self.openai_model_name = os.getenv('OPENAI_MODEL_NAME', "gpt-4")
         self.vision_tool_model = os.getenv('VISION_TOOL_MODEL', "CLAUDE_SONNET")
 
         self.ollama_url = os.getenv('OLLAMA_URL')
         self.qdrant_url = os.getenv('QDRANT_URL')
         self.browserless_url = os.getenv('BROWSERLESS_URL')
 
-
 config = Config()
 def load_config():
     global config
@@ -29,18 +28,22 @@ def load_config():
 def AGENT_MODEL_PROVIDER():
     return config.agent_model_provider
 
-def AGENT_MODEL():
-    return config.agent_model
+def AGENT_MODEL_NAME():
+    return config.agent_model_name
+
+def AGENT_MAX_ITER():
+    try:
+        int_val = int(config.agent_max_iter)
+        return int_val
+    except ValueError:
+        return 10
 
 def ANTHROPIC_KEY():
     return config.anthropic_key
 
 def OPENAI_API_KEY():
     return config.openai_api_key
 
-def OPENAI_MODEL_NAME():
-    return config.openai_model_name
-
 def OLLAMA_URL():
     return config.ollama_url
 
@@ -59,4 +62,3 @@ def BROWSERLESS_URL():
 def VISION_TOOL_MODEL():
     return config.vision_tool_model
 
-

src/dria_searching_agent/main.py

@@ -120,6 +120,7 @@ def __create_agents(self):
         for k,v in self.agents_config.items():
             agent = Agent(
                 **v,
+                max_iter=config.AGENT_MAX_ITER(), # TODO: maybe we can get this from agents.json
                 verbose=True,
                 llm = self.__get_model(),
                 tools=[
@@ -163,11 +164,11 @@ def __create_agents(self):
 
     def __get_model(self):
         if config.AGENT_MODEL_PROVIDER().lower() == "anthropic":
-            return ChatAnthropic(model=config.AGENT_MODEL(), api_key=config.ANTHROPIC_KEY())
+            return ChatAnthropic(model=config.AGENT_MODEL_NAME(), api_key=config.ANTHROPIC_KEY())
         elif config.AGENT_MODEL_PROVIDER().lower() == "openai":
-            return ChatOpenAI(api_key=config.AGENT_MODEL(), model=config.OPENAI_MODEL_NAME())
+            return ChatOpenAI(model=config.AGENT_MODEL_NAME(), api_key=config.OPENAI_API_KEY())
         elif config.AGENT_MODEL_PROVIDER().lower() == "ollama":
-            return ollama.Ollama(model=config.AGENT_MODEL(), base_url=config.OLLAMA_URL())
+            return ollama.Ollama(model=config.AGENT_MODEL_NAME(), base_url=config.OLLAMA_URL())
 
 _research_crew_instance = None
 def GetResearchCrew():

src/dria_searching_agent/server.py

@@ -30,7 +30,7 @@ def decorated_function(*args, **kwargs):
 
 @app.route('/search', methods=['POST'])
 @validate_query
-@limiter.limit("3 per minute")
+# @limiter.limit("3 per minute")
 def search():
     data = request.get_json()
     query = data.get('query')