Read the equation
Matrix multiplication sums the value vector at each key position weighted by that query’s attention probability. A 128-component value vector per head gives a 128-component output per head. This is a convex combination before subsequent dropout or projection when the supplied probabilities are nonnegative and sum to one.
Terms & definitions
| Term | Meaning in this source |
|---|---|
| probs | Attention probabilities |
| v | Value vectors |
How it works
Read the definition from its right-hand expression to the quantity on the left. Its surrounding source records the intended input, state, or parameter context. Where the source leaves an operation unspecified, this page preserves that gap.
Assumptions & limits
Source recovery establishes that the expression appears in this record. Its mathematical validity, implementation behavior, and interpretation must be assessed separately. See the linked source and research notes for the checks available for this version.
Original context
Read the surrounding source passage
k = concat(k_nope, k_rope) -> 192 per head (k_rope broadcast to all heads)
scores = q @ k^T * 0.135237
probs = softmax(scores, dim=-1) computed in float32, cast back to q.dtype
out = probs @ v -> 128 per head
out = o_proj(out.reshape(T, 16384)) -> 7168 fp8 (7168, 16384)
WHAT IS CACHED - THE POINT OF MLA
A conventional 128-head KV cache would store K and V per head:Exact original notation
out = probs @ v -> 128 per head