Curated · Fill in Blank
Top 25 Fill in Blank Interview Questions
Here are the top 25 AI interview questions in Fill in Blank format, sorted by community quality. Each entry links to a detailed answer with explanations, hints, and source-grade follow-ups. See all in this format →
Sorted by community quality · refreshed as new questions are published.
- 01Fill in the scaled dot-product attention formula.Fill in BlankEasy·Qual 4.0
- 02Fill in the memory delta when a 7B weight tensor moves from fp32 to bf16Fill in BlankEasy·Qual 4.0
- 03FLOPs vs FLOP/s: fill in the counts versus rate distinction.Fill in BlankEasy·Qual 4.0
- 04Adam optimizer state per parameter: fill in the moment buffers and bytesFill in BlankEasy·Qual 4.0
- 05Each row of the post-softmax attention weight matrix corresponds to which side of the QK product?Fill in BlankEasy·Qual 4.0
- 06Compute the effective batch from per-device batch, accum steps, and GPU countFill in BlankEasy·Qual 4.0
- 07End-to-end latency: fill in the queue, prefill and decode components.Fill in BlankEasy·Qual 4.0
- 08Name the two properties softmax guarantees for every row of the attention weight matrixFill in BlankEasy·Qual 4.0
- 09Complete the canonical attention formula: Attention(Q, K, V) = ___(QKᵀ / ___) · VFill in BlankEasy·Qual 4.0
- 10Fill in the SFT JSONL: the three required fields of a typical chat example.Fill in BlankEasy·Qual 4.0
- 11Each decode step loads ___ from HBM, regardless of how many tokens have already been generatedFill in BlankEasy·Qual 4.0
- 12Fill in the DeepSpeed config keys that select a ZeRO stageFill in BlankEasy·Qual 4.0
- 13Fill in the IVF nlist rule of thumb.Fill in BlankEasy·Qual 4.0Pinecone
- 14Complete the definition: prompt engineering is the practice of designing the LLM's ___ to shape its output, without modifying model ___.Fill in BlankEasy·Qual 4.0
- 15Complete the RMSNorm formulaFill in BlankEasy·Qual 4.0
- 16Given Q of shape (B, n_heads, T, d_head), the per-head attention output before concatenation has shape ___.Fill in BlankEasy·Qual 4.0
- 17Complete the pre-norm transformer block recipeFill in BlankEasy·Qual 4.0Mistral AI
- 18Modern 7B-class LLMs, fill in typical head counts and the resulting per-head dimensionFill in BlankEasy·Qual 4.0
- 19If d_model = 768 and num_heads = 12, what is d_head?Fill in BlankEasy·Qual 4.0
- 20Fill the blank: for a single head, QK^T has shape (T_query, ___).Fill in BlankEasy·Qual 4.0
- 21Compute the LoRA trainable param count for a 4096x4096 projection at rank r=8Fill in BlankEasy·Qual 4.0
- 22Complete the JSON-RPC 2.0 request structure used by MCPFill in BlankMedium·Qual 4.0
- 23Complete: Cohen's kappa >___ is 'substantial' agreement; >___ is 'almost perfect' agreementFill in BlankMedium·Qual 4.0
- 24Fill in the blanks: the special tokens that control sequence boundaries and conversation structure in modern LLMs.Fill in BlankMedium·Qual 4.0
- 25Fill in the blanks: query expansion / query rewriting in RAG and what it costs.Fill in BlankMedium·Qual 4.0
Related top picks
By experience level
Other formats
By topic