chore: refactor examples and clean CLI (#195)

deploy/examples/llm/configs/disaggregated/routerless_deployment.yaml → deploy/examples/llm/configs/disagg.yaml

@@ -16,22 +16,29 @@
 
 Frontend:
   model: deepseek-ai/DeepSeek-R1-Distill-Llama-8B
-  endpoint: dynamo-init.VllmWorkerRouterLess.generate
+  endpoint: dynamo.Processor.chat/completions
   port: 8000
 
-VllmWorkerRouterLess:
+Processor:
+  model: deepseek-ai/DeepSeek-R1-Distill-Llama-8B
+  block-size: 64
+  max-model-len: 16384
+  router: round-robin
+
+VllmWorker:
   model: deepseek-ai/DeepSeek-R1-Distill-Llama-8B
   enforce-eager: true
   kv-transfer-config: '{"kv_connector":"DynamoNixlConnector"}'
   block-size: 64
   max-model-len: 16384
   max-num-batched-tokens: 16384
-  remote-prefill: true
-  ServiceArgs:
-    workers: 1
-
+  conditional-disagg: true
+  tensor-parallel-size: 1
+  remote_prefill: true
+  max_local_prefill_length: 10
 
-PrefillWorkerRouterLess:
+# TODO - set all of these but model as default
+PrefillWorker:
   model: deepseek-ai/DeepSeek-R1-Distill-Llama-8B
   enforce-eager: true
   kv-transfer-config: '{"kv_connector":"DynamoNixlConnector"}'
@@ -39,5 +46,3 @@ PrefillWorkerRouterLess:
   max-model-len: 16384
   max-num-batched-tokens: 16384
   cuda-visible-device-offset: 1
-  ServiceArgs:
-    workers: 1

deploy/examples/llm/configs/disaggregated/router_based_deployment.yaml → deploy/examples/llm/configs/disagg_router.yaml

@@ -15,7 +15,7 @@
 
 Frontend:
   model: deepseek-ai/DeepSeek-R1-Distill-Llama-8B
-  endpoint: dynamo-init.Processor.chat/completions
+  endpoint: dynamo.Processor.chat/completions
   port: 8000
 
 Processor:

deploy/examples/llm/configs/monolith/routerless_deployment.yaml

-# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
-# SPDX-License-Identifier: Apache-2.0
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-# http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-Frontend:
-  model: deepseek-ai/DeepSeek-R1-Distill-Llama-8B
-  endpoint: dynamo-init.VllmWorkerRouterLess.generate
-  port: 8000
-
-VllmWorkerRouterLess:
-  model: deepseek-ai/DeepSeek-R1-Distill-Llama-8B
-  enforce-eager: true
-  max-model-len: 16384
-  max-num-batched-tokens: 16384
-  block-size: 64
\ No newline at end of file

deploy/examples/llm/disaggregated/routerless_deployment.py

-# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
-# SPDX-License-Identifier: Apache-2.0
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-# http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-from components.frontend import Frontend
-from components.routerless.prefill_worker import PrefillWorkerRouterLess
-from components.routerless.worker import VllmWorkerRouterLess
-
-Frontend.link(VllmWorkerRouterLess).link(PrefillWorkerRouterLess)

deploy/examples/llm/monolith/router_based_deployment.py → deploy/examples/llm/graphs/agg_router.py

@@ -18,5 +18,4 @@ from components.kv_router import Router
 from components.processor import Processor
 from components.worker import VllmWorker
 
-# Monolith with Router
 Frontend.link(Processor).link(Router).link(VllmWorker)

deploy/examples/llm/monolith/routerless_deployment.py → deploy/examples/llm/graphs/disagg.py

@@ -14,7 +14,8 @@
 # limitations under the License.
 
 from components.frontend import Frontend
-from components.routerless.worker import VllmWorkerRouterLess
+from components.prefill_worker import PrefillWorker
+from components.processor import Processor
+from components.worker import VllmWorker
 
-# Monolith without Router
-Frontend.link(VllmWorkerRouterLess)
+Frontend.link(Processor).link(VllmWorker).link(PrefillWorker)