Multiple Choice

Pick the formula that defines head_dim from hidden_size and num_heads.

Llama-2 70B uses 64 query heads, how many KV heads does it actually keep?

How does torch.nn.functional.scaled_dot_product_attention relate to FlashAttention?

Does FlashAttention change the attention output relative to standard attention?

Flash-Decoding parallelizes an axis FlashAttention v2 left alone. Pick which one.

Why is causal attention cheaper than full bidirectional at the same length?

Encoder self-attention versus decoder self-attention, what is the single structural difference at the mask level?

Inside a T5-style encoder-decoder, which sub-layer reads the encoder outputs, and through which projection?

Walk through how a token's d_model embedding is split into per-head pieces inside multi-head attention.

Bi-encoders vs cross-encoders, where does attention cross the query and document?

Identify the context vector inside scaled dot-product attention and what produces it.

BERT prepends a [CLS] token to every input, what is its actual job in the model?

What problem does chunked prefill solve in production LLM serving?

Name the additive bias term in attention and the two things it most commonly carries.

Attention temperature divides QK^T; sampling temperature divides output logits, distinguishwhere each lives.

After W_O projects the multi-head output back to d_model, what happens before the next sub-layer runs?

Injecting noise into pre-softmax attention scores, goal and risk?

Contrast ALiBi's linear distance bias with RoPE's vector rotation, which one modifies scores additively, and which one modifies Q and K directly?

Pick the statement that best describes a scoring rubric in LLM evaluation.

Pick the best definition: in LLM evaluation, a metric is _____.

Why might F1 score be a better choice than raw accuracy for evaluating entity extraction?

You have no reference answers for your eval set. Does that mean you cannot evaluate the model?

Offline eval now or wait for the A/B test results? Pick the key difference between online and offline evaluation.

For a machine translation eval, a senior engineer says BLEU is fine as the primary metric. When are they right, and when should you push back?

ROUGE-L, BERTScore, and LLM-as-judge are on the table for a summarization eval. Which covers what, and which drops first if budget is tight?