Idanwo ti o sopọ mọ itọsọna · Alabọde Ipele

Idanwo Backpropagation

Test backpropagation gradients, reverse dependency order, cached forward values, chain-rule multiplication, shared-path sums, and optimizer separation.

Jẹmọ itọsọna awọn ọnaItankalẹ afẹyinti
Ibeere 1 ti 8

What quantity does backpropagation compute for a neural network's relevant weights?

Awotẹlẹ gbogbo ibeere8 ibeere

Awọn idahun ati awọn alaye han bi o ṣe n mu ibeere naa.

  1. What quantity does backpropagation compute for a neural network's relevant weights?
    • Derivatives of the loss with respect to those weights
    • A causal percentage of total error uniquely assigned to each weight
    • The architecture drawing without any calculation
    • A permanently updated set of weights without an optimizer
  2. Ofin mathematiki wo ni o wa ni okan ti isọdọtun?
    • The Pythagorean Theorem
    • Bayes 'ero
    • Ofin pq ti iṣiro
    • Ilana kuadiratiki
  3. Why can reverse-mode backpropagation compute many parameter derivatives efficiently?
    • It avoids all local derivatives
    • It reuses graph structure and relevant intermediate results
    • It calculates only the output prediction again
    • It guarantees every implementation has exactly one forward pass of runtime
  4. Ni aṣẹ wo ni isọdọtun ṣe iṣiro awọn gradients?
    • Lati Layer igbewọle si ọna Layer ti o wu jade
    • Gbogbo awọn fẹlẹfẹlẹ nigbakanna laisi aṣẹ
    • Nikan fun arin farasin Layer
    • Lati awọn ti o wu Layer sẹhin si ọna kikọ sii Layer
  5. Does computing backpropagation gradients by itself establish that model weights were updated?
    • Yes; every backward computation automatically applies a learning rate
    • Yes; gradient values always replace the weights directly
    • No; the optimizer update is a separate step
    • Only if the network has no parameters
  6. In a conventional cached-backpropagation workflow, what does the forward computation provide for later derivative calculations?
    • Relevant intermediate values that can be retained for the backward computation
    • A completed optimizer update without any update rule
    • A guarantee that all later evaluation errors disappear
    • A permanent deletion of the training objective
  7. On one composed backpropagation path, local scalar derivatives are three and four. What is their chain-rule product?
    • Meje
    • Ọkan
    • Three quarters
    • Mejila
  8. A shared parameter has loss-gradient contributions two and five through two paths. What combined contribution is accumulated?
    • Mẹwa
    • Meje
    • Mẹta
    • Two fifths