AIF-C01 Fundamentals of AI and ML — All Questions With Answers

Refer to the exhibit.

```json
{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": [
        "s3:GetObject",
        "s3:PutObject"
      ],
      "Resource": "arn:aws:s3:::my-bucket/training/*"
    }
  ]
}
```

Refer to the exhibit.

```
2023-09-15 10:15:30,123 INFO     - Training job started
2023-09-15 10:15:35,456 INFO     - Epoch 1/10: loss=2.3456, accuracy=0.1234
2023-09-15 10:15:40,789 INFO     - Epoch 2/10: loss=2.3001, accuracy=0.1300
2023-09-15 10:15:46,012 INFO     - Epoch 3/10: loss=2.2800, accuracy=0.1350
2023-09-15 10:15:51,234 INFO     - Epoch 4/10: loss=2.3100, accuracy=0.1280
2023-09-15 10:15:56,456 WARNING - Loss increased from 2.2800 to 2.3100
2023-09-15 10:16:01,678 INFO     - Epoch 5/10: loss=2.3500, accuracy=0.1200
```

{
  "TrainingJobName": "my-training-job-1",
  "TrainingJobStatus": "Failed",
  "FailureReason": "AlgorithmError: OutOfMemoryError: CUDA out of memory. Tried to allocate 4.00 GiB (GPU 0; 8.00 GiB total capacity; 3.95 GiB already allocated; 2.50 GiB free; 4.00 GiB reserved in total by PyTorch)"
}

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": "s3:GetObject",
      "Resource": "arn:aws:s3:::my-training-data/*"
    }
  ]
}

Refer to the exhibit: CloudWatch Logs excerpt from a SageMaker real-time endpoint with instance type ml.m5.large. Average inference time is 500ms but increases over time. Logs show repeated 'MemoryError: cannot allocate memory' after several hours of operation.

Refer to the exhibit.

```json
{
  "TrainingJobStatus": "Failed",
  "FailureReason": "ClientError: ValidationException: The instance type ml.p3.2xlarge is not supported in the requested region (us-east-1)."
}
```

This is the output of `aws sagemaker describe-training-job --training-job-name my-training-job`.

Refer to the exhibit.

```yaml
Resources:
  MyNotebookInstance:
    Type: AWS::SageMaker::NotebookInstance
    Properties:
      InstanceType: ml.t2.medium
      VolumeSizeInGB: 5
      DirectInternetAccess: Enabled
```

Excerpt from an AWS CloudFormation template.

Refer to the exhibit.

```json
{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": "sagemaker:CreateTrainingJob",
      "Resource": "*"
    },
    {
      "Effect": "Allow",
      "Action": "s3:GetObject",
      "Resource": "arn:aws:s3:::my-bucket/*"
    }
  ]
}
```

Excerpt from an IAM policy attached to a SageMaker execution role.

{
  "ModelName": "my-xgboost-model",
  "TrainingJobName": "my-training-job",
  "TrainingJobStatus": "Completed",
  "SecondaryStatusTransitions": [
    {"Status": "Starting", "StartTime": "2024-01-01T10:00:00Z"},
    {"Status": "Training", "StartTime": "2024-01-01T10:05:00Z"},
    {"Status": "Completed", "StartTime": "2024-01-01T10:35:00Z"}
  ],
  "FinalMetricData": [
    {"MetricName": "validation:logloss", "Value": 0.45},
    {"MetricName": "validation:auc", "Value": 0.82}
  ],
  "ResourceConfig": {
    "InstanceType": "ml.m5.large",
    "InstanceCount": 1,
    "VolumeSizeInGB": 50
  }
}