Free H13-323_V1.0 – HCIP-AI Solution Architect V1.0 Practice Test Question

Select an option, then click Submit answer.

• ○ Self-attention mechanism in layers
• ○ Recurrent hidden state loop
• ○ Convolution filter weight sharing
• ○ Pooling window down sampling

Select an option, then click Submit answer.

• ○ Small labelled test datasets
• ○ Massive unlabelled text corpora
• ○ Manually written rule bases
• ○ Single domain image sets

Select an option, then click Submit answer.

• ○ Gradient clipping during training
• ○ Dropout masks between layers
• ○ Positional encoding added embeddings
• ○ Weight decay regularisation term

Select an option, then click Submit answer.

• ○ Classify input image regions
• ○ Compress model weight files
• ○ Retrieve documents from indexes
• ○ Generate tokens autoregressively step-by-step

Select an option, then click Submit answer.

• ○ Square root of key dimension
• ○ Total number of layers
• ○ Batch size of training
• ○ Vocabulary size of model

Select an option, then click Submit answer.

• ○ Duplicate every model layer
• ○ Route tokens to selected experts
• ○ Remove all feedforward networks
• ○ Convert text into images

Select all that apply, then click Submit answer.

• ○ Multi-head attention mechanism block
• ○ Feedforward network sublayer block
• ○ Recurrent memory cell block
• ○ Convolution pooling layer block

Select all that apply, then click Submit answer.

• ○ Learning rate training parameter
• ○ Temperature sampling parameter setting
• ○ Top-k sampling candidate limit
• ○ Weight decay regularisation coefficient

Select all that apply, then click Submit answer.

• ○ Compiling into OM format
• ○ Serving behind API gateway
• ○ Pre-training on general corpora
• ○ Fine-tuning on task data

Answer is in the explanation below.

Answer is in the explanation below.

Answer is in the explanation below.