Prueba de decodificación de capas intermedias y lentes Logit

Las respuestas y explicaciones aparecen a medida que realiza el cuestionario.

1. What readout does a plain logit lens reuse to obtain vocabulary scores from an intermediate state?
   • A required new image classifier for every layer
   • The model's appropriate final normalization and unembedding output projection
   • A random dictionary sort with no model parameters
   • A separate retrained tokenizer for each prompt
2. Why can the same output projection be applied mathematically at multiple residual-stream depths?
   • Every layer must contain an identical copy of the final state
   • The entire transformer is a single linear mapping
   • All intermediate token scores are guaranteed identical
   • The states have compatible hidden dimensions, although their representations may not be equally calibrated for that readout
3. ¿Qué problema soluciona la 'lente sintonizada' en relación con la lente logit simple?
   • Corrige la discrepancia por capa con una transformación afín aprendida
   • Elimina la necesidad de desincrustar
   • Hace que el modelo sea más rápido en la inferencia.
   • Reemplaza softmax con argmax
4. If an intermediate lens readout favors a correct factual token, what has been directly observed?
   • The unique causal location of the model's factual knowledge
   • A guarantee that all later output will be correct
   • That the particular probe maps that state to a high score for the token
   • The exact reasoning process of a conscious system
5. ¿Para qué es más útil una lente logit?
   • Comprimir pesos de modelos
   • Traducir entre idiomas
   • Activaciones de modelos de cifrado
   • Observando cómo y dónde se forma una predicción en profundidad
6. Why can a plain final-head readout be misleading on early transformer states?
   • Early states always use a different vocabulary file
   • Their representation distributions can differ from the final state for which the head was trained
   • Every early-layer vector is necessarily unstructured noise
   • No matrix operation can be applied before the final layer
7. A lens uses a model with a fifty-thousand-entry vocabulary and a 768-dimensional hidden state. How many vocabulary logits does its unembedding produce?
   • 768, one score per hidden coordinate only
   • One, because the token must be chosen before projection
   • Fifty thousand, one score per vocabulary entry
   • Exactly the number of transformer layers
8. What does softmax add to a logit-lens vocabulary-score display?
   • A normalized distribution over the projected vocabulary scores
   • Proof that every displayed probability is a calibrated belief
   • A causal intervention on the model's hidden states
   • A replacement for the output projection's vocabulary dimension