Maswali yanayounganishwa na mwongozo · Ngumu Kiwango

Sparse Autoencoder kwa Maswali ya Uchimbaji wa Kipengele

Soma kamusi chache za vipengele na uundaji upya, ukitumia jaribio la Daraja la Dhahabu, lango zilizojaa kupita kiasi, usukani na mipaka ya kielelezo.

Njia za mwongozo zinazohusianaSparse Autoencoder Kwa Uchimbaji wa Kipengele
Swali 1 ya 8

Which interpretability difficulty motivates sparse dictionary learning for neural activations?

Hakiki kila swali8 maswali

Majibu na maelezo huonekana unapojibu maswali.

  1. Which interpretability difficulty motivates sparse dictionary learning for neural activations?
    • Every model activation is already a complete dictionary definition
    • Several learned signals can overlap in an activation space
    • The tokenizer must be permanently removed
    • All neural weights are stored as image pixels
  2. What does an SAE sparsity constraint encourage for a given input?
    • Every feature to activate equally on every input
    • A mandatory removal of its decoder
    • A guarantee that every feature has exactly one perfect human label
    • Only a limited subset of learned feature activations to be nonzero
  3. Katika kazi ya Anthropic ya 'Scaling Monosemanticity', ni kipengele gani cha mfano maarufu?
    • Kipengele cha Golden Gate Bridge
    • Kipengele cha nambari saba
    • Kipengele cha rangi ya bluu pekee
    • Kipengele cha vokali
  4. What does an overcomplete SAE dictionary mean?
    • It must activate every feature for every token
    • It contains no reconstruction mapping
    • It has more learned feature directions than input activation dimensions
    • It always requires less storage than the input vector
  5. What does a monosemantic feature label aim to describe in this research context?
    • A guarantee that the feature never responds to related examples
    • A feature associated with one interpretable concept
    • A language model that has only one parameter
    • A token represented by exactly one raw byte
  6. SAEs zinawezaje kutumika kushawishi tabia ya modeli?
    • Kwa kubana kipengele kuwasha au kuzima ili kuelekeza matokeo
    • Kwa kufundisha tena mfano mzima kutoka mwanzo
    • Kwa kufuta tokenizer
    • Kwa kuibadilisha kuwa mfano wa picha
  7. What does an SAE’s reconstruction residual measure for an activation vector?
    • The difference between two unrelated document titles
    • The total number of words in the model vocabulary
    • Only the model’s hardware temperature
    • The difference between the original activation and its decoded reconstruction
  8. When an SAE is fitted to collected activations of a frozen language model, which weights need to be trained for that fitting?
    • Every original language-model weight must be updated
    • Only the tokenizer’s character list
    • The SAE’s own encoder and decoder weights
    • No parameters of the SAE can change