1Z0-1127 Fundamentals of Large Language Models — All Questions With Answers

Refer to the exhibit.

```
oci ai language text-classification --text "The product is amazing!"
{
  "data": {
    "labels": [
      {"name": "positive", "score": 0.98},
      {"name": "negative", "score": 0.01},
      {"name": "neutral", "score": 0.01}
    ]
  }
}
```

{
  "compartmentId": "ocid1.compartment.oc1..example",
  "definedTags": {},
  "description": "Allow group GenAIUsers to call generative-ai-family in compartment",
  "freeformTags": {},
  "id": "ocid1.policy.oc1..example",
  "name": "genai-policy",
  "state": "ACTIVE",
  "statements": [
    "Allow group GenAIUsers to use generative-ai-family in compartment ExampleCompartment"
  ]
}

Refer to the exhibit.

# OCI CLI output from `oci generative-ai model list`
{
  "data": [
    {
      "id": "ocid1.generativeaimodel.oc1..aaaaaa...",
      "model-name": "cohere.command",
      "capabilities": ["chat", "text-generation"],
      "context-window": 4096
    },
    {
      "id": "ocid1.generativeaimodel.oc1..bbbbbb...",
      "model-name": "cohere.base",
      "capabilities": ["text-generation"],
      "context-window": 8192
    },
    {
      "id": "ocid1.generativeaimodel.oc1..cccccc...",
      "model-name": "cohere.embed",
      "capabilities": ["embeddings"],
      "context-window": null
    }
  ]
}

Refer to the exhibit.

# OCI IAM policy for accessing generative AI models
Allow group GenAIDevelopers to use generative-ai-model-family in compartment ProdCompartment
Allow group GenAIDevelopers to read generative-ai-model in compartment ProdCompartment
Allow group GenAIDevelopers to create generative-ai-inference in compartment ProdCompartment

Refer to the exhibit.

# Error from OCI Generative AI inference API
{
  "code": "InferenceRequestTooLarge",
  "message": "The prompt plus max_new_tokens exceeds the model's context length of 8192 tokens.",
  "prompt_length": 6000,
  "max_new_tokens": 4000
}

IAM Policy:
Allow group GenAI-Users to use generative-ai-family in compartment ABC where request.operation = 'TextGeneration'

{
  "statement": "Allow group GenAIGroup to read generative-ai-models in compartment Production",
  "resource": "oci-generativeai:model",
  "compartment": "Production"
}

oci generative-ai inference chat \
--endpoint https://inference.generativeai.us-chicago-1.oci.oraclecloud.com \
--model-id "cohere.command" \
--compartment-id "ocid1.compartment.oc1..example" \
--messages '[{"role":"user","content":"Explain quantum computing."}]' \
--temperature 0.7 \
--top-p 0.9 \
--max-tokens 500

Error:
{
  "code": "InvalidParameter",
  "message": "The specified model 'my-custom-model' does not exist or you are not authorized to access it."
}

Refer to the exhibit.

```
$ oci ai language text-classification --text "I love this product!" --endpoint https://language.oci.oraclecloud.com
{
  "data": {
    "text": "I love this product!",
    "document-classification": [
      {
        "text": "I love this product!",
        "labels": [
          {
            "name": "Positive",
            "score": 0.98
          },
          {
            "name": "Negative",
            "score": 0.01
          },
          {
            "name": "Neutral",
            "score": 0.01
          }
        ],
        "version": "1.0"
      }
    ]
  }
}
```

Refer to the exhibit.

```json
{
  "compartmentId": "ocid1.compartment.oc1..example",
  "modelId": "cohere.command",
  "messages": [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "What is the capital of France?"}
  ],
  "maxTokens": 100,
  "temperature": 0.7,
  "topP": 0.9,
  "frequencyPenalty": 0.0,
  "presencePenalty": 0.0
}
```

Refer to the exhibit.

```
$ oci ai language fine-tuning-job create \
  --compartment-id ocid1.compartment.oc1..example \
  --model-id cohere.command \
  --training-dataset-uri "oci://bucket@namespace/train.jsonl" \
  --validation-dataset-uri "oci://bucket@namespace/val.jsonl" \
  --parameters '{"numEpochs": 5, "learningRate": 0.0001, "batchSize": 16}' \
  --max-duration-in-minutes 600
```

{
  "compartmentId": "ocid1.compartment.oc1..aaaaaa...",
  "servingMode": {
    "modelId": "ocid1.generativeaimodel.oc1..aaaaaa...",
    "servingType": "ON_DEMAND"
  },
  "inferenceRequest": {
    "prompt": "Explain the transformer architecture.",
    "maxTokens": 500,
    "temperature": 0.7,
    "topP": 0.9,
    "frequencyPenalty": 0.0,
    "presencePenalty": 0.0
  }
}