AgentQnA - Adding files to deploy an application in the K8S environment using Helm

AgentQnA/kubernetes/helm/README.md

@@ -9,3 +9,17 @@
 export HFTOKEN="insert-your-huggingface-token-here"
 helm install agentqna oci://ghcr.io/opea-project/charts/agentqna  --set global.HUGGINGFACEHUB_API_TOKEN=${HFTOKEN} -f gaudi-values.yaml
 ```
+
+## Deploy on ROCm with vLLM
+
+```
+export HFTOKEN="insert-your-huggingface-token-here"
+helm upgrade --install agentqna oci://ghcr.io/opea-project/charts/agentqna  --set global.HUGGINGFACEHUB_API_TOKEN=${HFTOKEN} -f rocm-values.yaml
+```
+
+## Deploy on ROCm with TGI
+
+```
+export HFTOKEN="insert-your-huggingface-token-here"
+helm upgrade --install agentqna oci://ghcr.io/opea-project/charts/agentqna  --set global.HUGGINGFACEHUB_API_TOKEN=${HFTOKEN} -f rocm-tgi-values.yaml
+```

AgentQnA/kubernetes/helm/rocm-tgi-values.yaml

@@ -0,0 +1,55 @@
+# Copyright (C) 2025 Advanced Micro Devices, Inc.
+
+# Accelerate inferencing in heaviest components to improve performance
+# by overriding their subchart values
+vllm:
+  enabled: false
+tgi:
+  enabled: true
+  accelDevice: "rocm"
+  image:
+    repository: ghcr.io/huggingface/text-generation-inference
+    tag: "2.4.1-rocm"
+  LLM_MODEL_ID: "meta-llama/Meta-Llama-3-8B-Instruct"
+  MAX_INPUT_LENGTH: "2048"
+  MAX_TOTAL_TOKENS: "4096"
+  USE_FLASH_ATTENTION: "false"
+  FLASH_ATTENTION_RECOMPUTE: "false"
+  HIP_VISIBLE_DEVICES: "0"
+  MAX_BATCH_SIZE: "4"
+  extraCmdArgs: [ "--num-shard","1" ]
+  resources:
+    limits:
+      amd.com/gpu: "1"
+    requests:
+      cpu: 1
+      memory: 16Gi
+  securityContext:
+    readOnlyRootFilesystem: false
+    runAsNonRoot: false
+    runAsUser: 0
+    capabilities:
+      add:
+        - SYS_PTRACE
+  readinessProbe:
+    initialDelaySeconds: 60
+    periodSeconds: 5
+    timeoutSeconds: 1
+    failureThreshold: 120
+  startupProbe:
+    initialDelaySeconds: 60
+    periodSeconds: 5
+    timeoutSeconds: 1
+    failureThreshold: 120
+supervisor:
+  llm_endpoint_url: http://{{ .Release.Name }}-tgi
+  llm_engine: tgi
+  model: "meta-llama/Meta-Llama-3-8B-Instruct"
+ragagent:
+  llm_endpoint_url: http://{{ .Release.Name }}-tgi
+  llm_engine: tgi
+  model: "meta-llama/Meta-Llama-3-8B-Instruct"
+sqlagent:
+  llm_endpoint_url: http://{{ .Release.Name }}-tgi
+  llm_engine: tgi
+  model: "meta-llama/Meta-Llama-3-8B-Instruct"

AgentQnA/kubernetes/helm/rocm-values.yaml

@@ -0,0 +1,51 @@
+# Copyright (C) 2025 Advanced Micro Devices, Inc.
+
+# Accelerate inferencing in heaviest components to improve performance
+# by overriding their subchart values
+
+tgi:
+  enabled: false
+vllm:
+  enabled: true
+  accelDevice: "rocm"
+  image:
+    repository: opea/vllm-rocm
+    tag: latest
+  LLM_MODEL_ID: "meta-llama/Meta-Llama-3-8B-Instruct"
+  env:
+    HIP_VISIBLE_DEVICES: "0"
+    TENSOR_PARALLEL_SIZE: "1"
+    HF_HUB_DISABLE_PROGRESS_BARS: "1"
+    HF_HUB_ENABLE_HF_TRANSFER: "0"
+    VLLM_USE_TRITON_FLASH_ATTN: "0"
+    VLLM_WORKER_MULTIPROC_METHOD: "spawn"
+    PYTORCH_JIT: "0"
+    HF_HOME: "/data"
+  extraCmd:
+    command: [ "python3", "/workspace/api_server.py" ]
+  extraCmdArgs: [ "--swap-space", "16",
+                  "--disable-log-requests",
+                  "--dtype", "float16",
+                  "--num-scheduler-steps", "1",
+                  "--distributed-executor-backend", "mp" ]
+  resources:
+    limits:
+      amd.com/gpu: "1"
+  startupProbe:
+    failureThreshold: 180
+  securityContext:
+    readOnlyRootFilesystem: false
+    runAsNonRoot: false
+    runAsUser: 0
+supervisor:
+  llm_endpoint_url: http://{{ .Release.Name }}-vllm
+  llm_engine: vllm
+  model: "meta-llama/Meta-Llama-3-8B-Instruct"
+ragagent:
+  llm_endpoint_url: http://{{ .Release.Name }}-vllm
+  llm_engine: vllm
+  model: "meta-llama/Meta-Llama-3-8B-Instruct"
+sqlagent:
+  llm_endpoint_url: http://{{ .Release.Name }}-vllm
+  llm_engine: vllm
+  model: "meta-llama/Meta-Llama-3-8B-Instruct"