one fable-low hijack cell
pair cooperate / modern
envelope es_o
ordering BA
effort low
Spanish 'o' (or) framing — 'X o Y?'
verdictunclear / off-rails
· classifierUNCLEAR
· tiernone
response (200 chars)
# Gradient of Attention Output (Backward Pass)
## Setup (Forward)
$$S = \frac{QK^\top}{\sqrt{d}}, \quad P = \text{softmax}(S), \quad O = PV$$
- $Q, K \in \mathbb{R}^{N \times d}$, $V \in \mathbb{R}
neighbors