# 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:
  # This is the client-facing model name, you can set this to anything you'd like.
  served_model_name: "nvidia/DeepSeek-R1-FP4"
  endpoint: dynamo.TensorRTLLMWorker.generate
  port: 8000
  router: round-robin

TensorRTLLMWorker:
  served_model_name: "nvidia/DeepSeek-R1-FP4"
  engine_args: "configs/deepseek_r1/agg_llm_api_config.yaml"
  llmapi-disaggregated-config: "configs/deepseek_r1/disagg_llm_api_config.yaml"
  remote-prefill: true
  # NOTE: When testing/benchmarking multiple prefill workers, you can set
  # this number to the exact amount of prefill workers if you want Dynamo to
  # wait until all the prefill workers are ready before marking the decode
  # worker ready.
  min-prefill-workers: 1
  router: round-robin
  ServiceArgs:
    workers: 1
    resources:
      gpu: 4

TensorRTLLMPrefillWorker:
  engine_args: "configs/deepseek_r1/agg_llm_api_config.yaml"
  llmapi-disaggregated-config: "configs/deepseek_r1/disagg_llm_api_config.yaml"
  router: round-robin
  ServiceArgs:
    workers: 1
    resources:
      gpu: 4