ai-model-distillation-for-financial-data/demo-values.yaml at main · NVIDIA-AI-Blueprints/ai-model-distillation-for-financial-data · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
##############################################################
# example values for use with tutorial workflow and minikube #
##############################################################

existingSecret: ngc-api
existingImagePullSecret: nvcrimagepullsecret
data-store:
  enabled: true
  external:
    rootUrl: http://data-store.test
    domain: data-store.test
  persistence:
    size: 2Gi
customizer:
  enabled: true
  modelsStorage:
    storageClassName: standard
  customizationTargets:
    overrideExistingTargets: true
    targets:
      meta/llama-3.2-1b-instruct@2.0:
        enabled: true
      meta/llama-3.2-3b-instruct@2.0:
        enabled: true
      meta/llama-3.1-8b-instruct@2.0:
        enabled: true
  # WandB secret configuration for customizer subchart
  wandb:
    secretValue: "" # Clear default value to use existingSecret instead
    existingSecret: "wandb-api-key"
    existingSecretKey: "api-key"
  customizerConfig:
    training:
      pvc:
        storageClass: "standard"
        volumeAccessMode: "ReadWriteOnce"
      # Configure training container with WANDB environment variables
      # Following NVIDIA NeMo documentation - container_defaults only supports direct values, not valueFrom
      # WANDB_API_KEY will be automatically injected by deploy-nmp.sh if set in ~/.env
      container_defaults:
        env:
          - name: WANDB_BASE_URL
            value: "https://api.wandb.ai"

    # Weights & Biases configuration
    # If wandb-api-key secret exists, training metrics will be logged to W&B
    wandb:
      entity: null # Optional: set your W&B team/username
      project: "ai-model-distillation-financial"
evaluator:
  milvus:
    enabled: false
guardrails:
  enabled: false
ingress:
  enabled: true
  annotations:
    nginx.ingress.kubernetes.io/proxy-body-size: 1024m
deployment-management:
  deployments:
    defaultStorageClass: standard