Flow Matching Quiz

Answers and explanations appear as you take the quiz.

1. What does a flow matching model directly learn to predict?
   • A velocity field that moves samples from noise toward data
   • A classification label for each image
   • The compressed JPEG bytes of an image
   • A fixed lookup table of pixel values
2. How are samples generated from a trained flow matching model?
   • By sampling a discrete Markov chain of words
   • By integrating the learned velocity field with an ODE solver starting from noise
   • By running gradient descent on the input image
   • By averaging the whole training set
3. What makes the conditional flow matching loss tractable to train?
   • It ignores the data entirely
   • It requires solving the full marginal field analytically
   • It conditions on a single data point and uses a simple known interpolation path
   • It only works on one image at a time forever
4. Why does the 'rectified flow' variant enable faster generation?
   • It uses larger neural networks
   • It skips training entirely
   • It converts images to text first
   • It straightens the probability paths so fewer ODE solver steps are needed
5. Which named image model’s paper explicitly describes a rectified-flow transformer?
   • Stable Diffusion 3
   • Flux, as established by the cited source set
   • ResNet-50
   • YOLOv8
6. In conditional flow matching, what is a typical interpolation between a noise sample and a data sample at time t?
   • A random shuffle of pixels
   • x_t = (1-t)*noise + t*data, a straight-line blend
   • Squaring every pixel value
   • Sorting pixels by brightness
7. In conditional flow matching, how is the marginal vector field related to conditional vector fields?
   • It is the largest vector in the training batch
   • It is a fixed class label independent of time
   • It is the conditional expectation of the conditional vector field over the conditioning variable
   • It is the average pixel value of the data
8. Beyond generative modeling, what other task does the original rectified-flow paper describe?
   • Speech diarization with speaker embeddings
   • Classifying handwritten digits from labels alone
   • Predicting calendar availability
   • Data or domain transfer between distributions