chore: update vLLM v0.19.0

.dockerignore

@@ -7,4 +7,4 @@ db/*
 caddy/caddy_config
 caddy/caddy_data
 tests/
-grafana/*
+grafana/*

.github/scripts/wait_for_ci_services.sh

@@ -9,12 +9,12 @@ ATTEMPT=1
 
 while [ $ATTEMPT -le $MAX_ATTEMPTS ]; do
     echo "Waiting for nilai to become healthy... API:[$API_HEALTH_STATUS] MODEL:[$MODEL_HEALTH_STATUS] NUC_API:[$NUC_API_HEALTH_STATUS] (Attempt $ATTEMPT/$MAX_ATTEMPTS)"
-    
+
     echo "===== Model Container Logs (last 50 lines) ====="
     docker logs --tail 50 nilai_gpt_20b_gpu_1 2>&1
     docker ps --format "table {{.Names}}\t{{.Status}}\t{{.Image}}"
     echo "================================================="
-    
+
     sleep 30
     API_HEALTH_STATUS=$(docker inspect --format='{{.State.Health.Status}}' nilai-api 2>/dev/null)
     MODEL_HEALTH_STATUS=$(docker inspect --format='{{.State.Health.Status}}' nilai_gpt_20b_gpu_1 2>/dev/null)

.pre-commit-config.yaml

@@ -15,20 +15,12 @@ repos:
       - id: ruff  # Runs Ruff's linter
         exclude: "packages/verifier/"  # Excludes the "packages/verifier/" directory from linting
 
-  # Second repository: Pyright for static type checking
-  - repo: local
-    hooks:
-      - id: pyright
-        name: pyright
-        entry: pyright
-        language: system
-        types: [python]
   # Third repository: Local hooks (custom pre-commit checks)
   - repo: local
     hooks:
       - id: pytest  # Defines a local hook for running pytest
         name: pytest  # A human-readable name for the hook
-        entry: pytest  # The command to execute
+        entry: uv run pytest  # The command to execute
         language: system  # Uses the system-installed pytest instead of a virtual environment
         types: [python]  # Applies only to Python files
         pass_filenames: false  # Run pytest on the entire test suite rather than just changed files

caddy/.gitignore

@@ -1,2 +1,2 @@
 caddy_config/
-caddy_data/
+caddy_data/

docker/api.Dockerfile

@@ -16,4 +16,4 @@ uv sync
 
 EXPOSE 8080
 
-CMD ["./launch.sh"]
+CMD ["./launch.sh"]

docker/vllm.Dockerfile

@@ -1,4 +1,4 @@
-FROM vllm/vllm-openai:v0.11.2
+FROM vllm/vllm-openai:v0.19.0
 
 # # Specify model name and path during build
 # ARG MODEL_NAME=llama_1b_cpu

nilai-api/alembic/README

@@ -1 +1 @@
-Generic single-database configuration with an async dbapi.
+Generic single-database configuration with an async dbapi.

nilai-api/src/nilai_api/config/auth.py

@@ -4,12 +4,12 @@
 
 class AuthConfig(BaseModel):
     auth_strategy: Literal["api_key", "jwt", "nuc"] = Field(
-        description="Authentication strategy"
+        default="api_key", description="Authentication strategy"
     )
     nilauth_trusted_root_issuers: List[str] = Field(
-        description="Trusted root issuers for nilauth"
+        default_factory=list, description="Trusted root issuers for nilauth"
     )
-    credit_api_token: str = Field(description="Credit service API token")
+    credit_api_token: str = Field(default="", description="Credit service API token")
     auth_token: Optional[str] = Field(
         default=None, description="Auth token for e2e tests and development"
     )

nilai-api/src/nilai_api/config/database.py

@@ -2,11 +2,11 @@
 
 
 class DatabaseConfig(BaseModel):
-    user: str = Field(description="Database user")
-    password: str = Field(description="Database password")
-    host: str = Field(description="Database host")
-    port: int = Field(description="Database port")
-    db: str = Field(description="Database name")
+    user: str = Field(default="", description="Database user")
+    password: str = Field(default="", description="Database password")
+    host: str = Field(default="", description="Database host")
+    port: int = Field(default=5432, description="Database port")
+    db: str = Field(default="", description="Database name")
 
 
 class DiscoveryConfig(BaseModel):
@@ -17,4 +17,6 @@ class DiscoveryConfig(BaseModel):
 
 
 class RedisConfig(BaseModel):
-    url: str = Field(description="Redis URL for rate limiting")
+    url: str = Field(
+        default="redis://localhost:6379", description="Redis URL for rate limiting"
+    )

nilai-api/src/nilai_api/config/nildb.py

@@ -4,11 +4,18 @@
 
 
 class NilDBConfig(BaseModel):
-    nilchain_url: str = Field(..., description="The URL of the Nilchain")
-    nilauth_url: str = Field(..., description="The URL of the Nilauth")
-    nodes: List[str] = Field(..., description="The URLs of the Nildb nodes")
-    builder_private_key: str = Field(..., description="The private key of the builder")
-    collection: Uuid = Field(..., description="The ID of the collection")
+    nilchain_url: str = Field(default="", description="The URL of the Nilchain")
+    nilauth_url: str = Field(default="", description="The URL of the Nilauth")
+    nodes: List[str] = Field(
+        default_factory=list, description="The URLs of the Nildb nodes"
+    )
+    builder_private_key: str = Field(
+        default="", description="The private key of the builder"
+    )
+    collection: Uuid = Field(
+        default=Uuid("00000000-0000-0000-0000-000000000000"),
+        description="The ID of the collection",
+    )
 
     @field_validator("nodes", mode="before")
     @classmethod

nilai-models/run.sh

@@ -30,4 +30,4 @@ main() {
     exit $?
 }
 
-main "$@"
+main "$@"

tests/e2e/test_chat_completions.py

@@ -625,10 +625,9 @@ def test_chat_completion_high_temperature(client):
     assert response, "High temperature request should return a valid response"
     assert response.choices, "Response should contain choices"
     assert len(response.choices) > 0, "At least one choice should be present"
-    assert (
-        response.choices[0].message.content
-        or response.choices[0].message.reasoning_content
-    ), "Response should contain content or reasoning_content"
+    assert response.choices[0].message.content or getattr(
+        response.choices[0].message, "reasoning", None
+    ), "Response should contain content or reasoning"
 
 
 def test_model_streaming_request_high_token(client):
@@ -689,10 +688,8 @@ def test_web_search(client, model, high_web_search_rate_limit):
     assert len(response.choices) > 0, "Response should contain at least one choice"
 
     content = response.choices[0].message.content
-    reasoning_content = getattr(response.choices[0].message, "reasoning_content", None)
-    assert content or reasoning_content, (
-        "Response should contain content or reasoning_content"
-    )
+    reasoning = getattr(response.choices[0].message, "reasoning", None)
+    assert content or reasoning, "Response should contain content or reasoning"
 
     sources = getattr(response, "sources", None)
     assert sources is not None, "Sources field should not be None"

tests/e2e/test_chat_completions_http.py

@@ -791,11 +791,9 @@ def test_web_search(client, model, high_web_search_rate_limit):
 
     message = response_json["choices"][0].get("message", {})
     content = message.get("content", "")
-    reasoning_content = message.get("reasoning_content", "")
+    reasoning = message.get("reasoning", "")
 
-    assert content or reasoning_content, (
-        "Response should contain content or reasoning_content"
-    )
+    assert content or reasoning, "Response should contain content or reasoning"
 
     sources = response_json.get("sources")
     if sources is not None: