quiz.json

{
  "lesson": "61-cross-attention-fusion",
  "title": "Cross-Attention Fusion",
  "questions": [
    {
      "stage": "pre",
      "question": "What is the operational difference between early fusion and late fusion in vision-language modeling?",
      "options": [
        "Early fusion runs first in wall-clock time",
        "Early fusion concatenates image and text tokens into one sequence; late fusion keeps them separate and bridges via cross-attention at each block",
        "Late fusion uses more memory",
        "Early fusion is illegal"
      ],
      "correct": 1,
      "explanation": "Chameleon and Emu3 are early-fusion; Flamingo and BLIP-2 are late-fusion. The architectural choice changes mask shapes and KV caching."
    },
    {
      "stage": "pre",
      "question": "In a decoder block with both self-attention and cross-attention, which one uses a causal mask?",
      "options": [
        "Both",
        "Self-attention only; the image is fully observed and cross-attention has no temporal order",
        "Cross-attention only",
        "Neither"
      ],
      "correct": 1,
      "explanation": "Text generation is autoregressive, so text self-attention is causal. Image tokens are all visible before any text is decoded."
    },
    {
      "stage": "check",
      "question": "Why is the cross-attention KV cache built once per image?",
      "options": [
        "PyTorch requires caching",
        "Image keys and values do not change as text is decoded, so projecting memory once and reusing the K and V across all decode steps is the whole inference speedup",
        "Caching saves disk space",
        "It avoids using the GPU"
      ],
      "correct": 1,
      "explanation": "The vision encoder runs once. Its KV projection is computed once. Each text token reuses the cache for free."
    },
    {
      "stage": "check",
      "question": "What is the output shape of cross-attention when text length is Nt=10 and image length is Nv=197 with hidden=256?",
      "options": [
        "(B, 197, 256)",
        "(B, 10, 256)",
        "(B, 10, 197)",
        "(B, 207, 256)"
      ],
      "correct": 1,
      "explanation": "Cross-attention output shape follows the query stream, so output is (B, Nt, hidden) regardless of key length."
    },
    {
      "stage": "check",
      "question": "Why is no mask used on cross-attention?",
      "options": [
        "Masks are too expensive",
        "The image is fully observed before text generation begins; every text position may attend to every patch",
        "PyTorch does not support cross masks",
        "Masks always cause NaN"
      ],
      "correct": 1,
      "explanation": "There is no temporal order on image patches and no leakage to prevent, so cross-attention sees the whole image."
    },
    {
      "stage": "post",
      "question": "Which extension would you add to recover Flamingo-style stability during training?",
      "options": [
        "Drop the self-attention layer",
        "Insert a learned tanh gate on the cross-attention residual so the model can start from text-only behavior and grow into the image stream",
        "Remove the FFN",
        "Use only one attention head"
      ],
      "correct": 1,
      "explanation": "Flamingo's tanh gate starts at zero, recovers text-only behavior at init, and gives the model a smooth ramp into cross-attention."
    }
  ]
}