Útmutatóhoz kapcsolódó kvíz · Közepes Szint
Logit objektív és köztes réteg dekódoló kvíz
Test intermediate residual readouts, normalization and unembedding, tuned affine probes, vocabulary logits, softmax, and limits of causal interpretation.
kérdés 1 -ból 8
What readout does a plain logit lens reuse to obtain vocabulary scores from an intermediate state?
Tekintse meg az összes kérdést8 kérdéseket
A válaszok és magyarázatok megjelennek a kvíz kitöltésekor.
- 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
- 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
- Milyen problémát old meg a "tuningolt lencse" a sima logit objektívhez képest?
- Megtanult affin transzformációval korrigálja a rétegenkénti eltérést
- Megszünteti a beágyazás megszüntetésének szükségességét
- Ezáltal a modell gyorsabban vonható le
- A softmax-ot argmax-ra cseréli
- 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
- Mire a legközvetlenebbül hasznos a logit objektív?
- Modellsúlyok tömörítése
- Nyelvek közötti fordítás
- Modellaktiválások titkosítása
- Figyeli, hogyan és hol alakul ki egy előrejelzés a mélységben
- 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
- 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
- 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