ray-embedding 0.13.18__tar.gz → 0.14.0__tar.gz

{ray_embedding-0.13.18 → ray_embedding-0.14.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ray-embedding
-Version: 0.13.18
+Version: 0.14.0
 Summary: Deploy SentenceTransformers embedding models to a ray cluster
 Author: Crispin Almodovar
 Author-email: 

{ray_embedding-0.13.18 → ray_embedding-0.14.0}/ray_embedding/deploy.py

@@ -1,4 +1,5 @@
 import os
+from typing import Any, Dict
 
 import torch
 from ray.serve import Application
@@ -6,7 +7,7 @@ from ray.serve import Application
 from ray_embedding.dto import AppConfig, ModelDeploymentConfig, DeployedModel, NodeReaperConfig
 from ray_embedding.embedding_model import EmbeddingModel
 from ray_embedding.node_reaper import NodeReaper, NODE_REAPER_DEPLOYMENT_NAME
-from ray_embedding.utils import node_affinity_for_head, node_affinity_for_worker, HEAD_NODE_IP
+from ray_embedding.utils import get_head_node_id
 from ray_embedding.model_router import ModelRouter
 
 
@@ -56,7 +57,7 @@ def build_app(args: AppConfig) -> Application:
 
     node_reaper_config = args.node_reaper or NodeReaperConfig()
 
-    node_reaper_kwargs = {
+    node_reaper_kwargs: Dict[str, Any] = {
         "ssh_user": node_reaper_config.ssh_user,
         "ssh_private_key": node_reaper_config.ssh_private_key,
     }
@@ -67,13 +68,20 @@ def build_app(args: AppConfig) -> Application:
 
     node_reaper = NodeReaper.options( 
         name=NODE_REAPER_DEPLOYMENT_NAME,
-        ray_actor_options={"num_cpus": 0.25, f"node:{HEAD_NODE_IP}": 1},
+        ray_actor_options={"num_cpus": 0.25, "resources": {"node_type:head": 1}},
         autoscaling_config={"initial_replicas": 1, "min_replicas": 1, "max_replicas": 1}
     ).bind(**node_reaper_kwargs)
 
     deployed_models = {model_config.served_model_name: build_model(model_config, node_reaper) for model_config in models}
+    model_router_kwargs = {
+        "deployed_models": deployed_models,
+        "path_prefix": model_router.path_prefix,
+        "max_concurrency": model_router.max_concurrency,
+        "node_reaper": node_reaper
+    }
     router = ModelRouter.options(
-        name=model_router.deployment
-    ).bind(deployed_models, model_router.path_prefix, node_reaper)
+        name=model_router.deployment,
+        ray_actor_options={"num_cpus": 0.25, "resources": {"node_type:worker": 1}}
+    ).bind(**model_router_kwargs)
     
     return router

{ray_embedding-0.13.18 → ray_embedding-0.14.0}/ray_embedding/utils.py

@@ -21,17 +21,6 @@ def get_head_node_id() -> Tuple[str, str]:
         raise RuntimeError("Unable to locate the head node ID for NodeReaper deployment.") from exc
 
 
-HEAD_NODE_ID, HEAD_NODE_IP = get_head_node_id()
-
-
-def node_affinity_for_head() -> NodeAffinitySchedulingStrategy:
-    return NodeAffinitySchedulingStrategy(node_id=HEAD_NODE_ID, soft=False)
-
-
-def node_affinity_for_worker() -> NodeAffinitySchedulingStrategy:
-    return NodeAffinitySchedulingStrategy(node_id=NotIn(HEAD_NODE_ID), soft=False)
-
-
 def get_node_reaper_handle() -> DeploymentHandle:
     try:
         return serve.context.get_deployment_handle(NODE_REAPER_DEPLOYMENT_NAME)

{ray_embedding-0.13.18 → ray_embedding-0.14.0}/ray_embedding.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ray-embedding
-Version: 0.13.18
+Version: 0.14.0
 Summary: Deploy SentenceTransformers embedding models to a ray cluster
 Author: Crispin Almodovar
 Author-email: 

{ray_embedding-0.13.18 → ray_embedding-0.14.0}/setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = ray-embedding
-version = 0.13.18
+version = 0.14.0
 author = Crispin Almodovar
 author_email = 
 description = Deploy SentenceTransformers embedding models to a ray cluster