[BUG]: NIM helm chart templates missing model.profiles configuration for embedding and reranking NIMs

[mbelur]

Version

2.6.0

Describe the bug.

The Helm chart templates for embedding and reranking NIMs do not pass through the model.profiles configuration to the NIMCache custom resource, causing the NIM Operator to download all available profiles instead of only the user-specified profile.

Expected Behavior:
When a user specifies a single profile in their values override file:

  nimOperator:
    nvidia-nim-llama-nemotron-embed-vl-1b-v2:
      model:
        profiles:
        - f7391ddbcb95b2406853526b8e489fedf20083a2420563ca3e65358ff417b10f

Only that profile should be downloaded by the NIMCache job.

Actual Behavior:

All available profiles are downloaded, consuming:

• Significantly more storage
• Longer cache job execution time
• Unnecessary network bandwidth

This impacts deployment in resource-constrained environment where storage is a critical `factor.```

Minimum reproducible example

Step 1: Create minimal values override file (test-values.yaml):

  imagePullSecret:
    name: "ngc-secret"
    create: true
    password: "<YOUR_NGC_API_KEY>"

  ngcApiSecret:
    name: "ngc-api"
    create: true
    password: "<YOUR_NGC_API_KEY>"

  nimOperator:
    nim-llm:
      enabled: false

    nvidia-nim-llama-nemotron-embed-1b-v2:
      enabled: false

    nvidia-nim-llama-nemotron-embed-vl-1b-v2:
      enabled: true
      model:
        profiles:
          - f7391ddbcb95b2406853526b8e489fedf20083a2420563ca3e65358ff417b10f
    nvidia-nim-llama-nemotron-rerank-1b-v2:
      enabled: false

  Step 2: Deploy the chart:
  
    helm upgrade --install rag -n test-nim-bug https://helm.ngc.nvidia.com/nvidia/blueprint/charts/nvidia-blueprint-rag-v2.6.0.tgz \
--username '$oauthtoken' \
--password "${NGC_API_KEY}" \
-f test-values.yaml \
--create-namespace


  Step 3: Verify the issue:

  Check the NIMCache spec (should have model.profiles but doesn't):
  kubectl get nimcache nemotron-vlm-embedding-ms-cache -n test-nim-bug -o yaml | grep -A 10 "spec:"

  Expected output:
  spec:
    source:
      ngc:
        modelPuller: "nvcr.io/nim/nvidia/llama-nemotron-embed-vl-1b-v2:1.12.0"
        pullSecret: ngc-secret
        authSecret: ngc-api
        model:
          profiles:
            - f7391ddbcb95b2406853526b8e489fedf20083a2420563ca3e65358ff417b10f

  Actual output:
  spec:
    source:
      ngc:
        modelPuller: "nvcr.io/nim/nvidia/llama-nemotron-embed-vl-1b-v2:1.12.0"
        pullSecret: ngc-secret
        authSecret: ngc-api
        # model section is missing
   
  Check how many profiles are being downloaded:
  kubectl get job -n test-nim-bug nemotron-vlm-embedding-ms-cache-job \
    -o jsonpath='{.spec.template.spec.containers[0].args}' | jq

  Expected: Single profile ID in args
  ["--profiles", "f7391ddbcb95b2406853526b8e489fedf20083a2420563ca3e65358ff417b10f"]

  Actual: All 11 profile IDs
  [
    "--profiles",
    "30deb3f6ab82ce062835b4e97894ad0ea76b5459ecd7f1f113ee7f4a920f81e9",
    "448d9749eede748c122f71cb0d0ad009557b5f6d7c1599cce7dfe24af4474d83",
    "44d1f6e283eb0365efb617f8c255e42f9d8c97fe039774c2837cdfcef54296ce",
    "8f04bf773931e5b983e6812a1bae11bbccf68f4cb62644421c42c3174c0ae143",
    "91b1800959ef14ba7340038e06e3ac51d7fa71975ba6e89c580affaf3cce7788",
    "a1c13d988e661ffcff9de9a919de942e2257c8a1990280e18f1fcfbf1bdd39e5",
    "a2a98d9d648d970f17ebca4b8ddacf96aaf7266ec66c74c83c6b42e3f5f5b97c",
    "ac9898291cd6590f16956c43779ee8a191bd9ff1c091cf2ddecf51ea13f72877",
    "c10e0d36e7605cdd0ccbb0e25f0be6d97c00b4113cc9242b577d550b92d83da8",
    "c3caccfe9fcbc6c3d8d911cfa19992e6bf3398d38298cfa97da79ad8f5f35723",
    "f7391ddbcb95b2406853526b8e489fedf20083a2420563ca3e65358ff417b10f"
  ]

Relevant log output

Full env printout

Other/Misc.

No response

Code of Conduct

• I agree to follow THIS PROJECT's Code of Conduct
• I have searched the open bugs and have found no duplicates for this bug report