Questionário vinculado ao guia · Difícil Nível

Questionário sobre vetores de funções e representações de tarefas

Estude como os vetores de função são extraídos de cabeças de atenção, injetados em estados ocultos e testados em contextos de tarefas.

Caminhos de guia relacionadosVetores de funções e representações de tarefas
Pergunta 1 de 8

What does a function vector represent in the studied language models?

Visualize cada pergunta8 perguntas

As respostas e explicações aparecem conforme você responde ao teste.

  1. What does a function vector represent in the studied language models?
    • A compact activation-space representation associated with a demonstrated task
    • The model’s full weight matrix
    • A single token embedding in every setting
    • The optimizer’s learning rate
  2. Como um vetor de função é normalmente construído?
    • Somando todas as incorporações de palavras no vocabulário
    • Calculando a média dos resultados de cabeças de atenção que transportam tarefas específicas sobre prompts de exemplo
    • Ajustando o modelo em um exemplo
    • Lendo as probabilidades softmax finais
  3. What did the function-vector paper observe after injecting a vector into a zero-shot prompt?
    • The model always crashed
    • It guaranteed correct answers on any task
    • It could trigger the associated task in the tested settings
    • It permanently changed the model weights
  4. Qual técnica de análise sustenta a localização das cabeças que carregam um vetor de função?
    • Descida gradiente na perda
    • Pesquisa de feixe
    • Regularização de abandono
    • Análise de mediação causal
  5. What do function-vector experiments provide evidence for?
    • Some models form reusable internal representations associated with tasks
    • All language models only memorize text
    • Every model has one universal task vector
    • Tokenization alone determines every behavior
  6. Why are function vectors interesting to steering research?
    • They guarantee safe or correct behavior
    • They provide an activation-level intervention that can influence task behavior in tested settings
    • They eliminate prompts for every application
    • They make all model decisions fully transparent
  7. At what part of the transformer did the paper report strong function-vector effects?
    • Only the input embedding table
    • Only the final output logits
    • Middle layers in the tested models
    • O vocabulário do tokenizador
  8. What happened when researchers summed some function vectors?
    • The vectors always canceled to zero
    • The model’s weights were permanently updated
    • Every possible task was composed perfectly
    • Some combinations triggered new complex tasks in the tested experiments