Provedllmlarge-language-modelsformal-proofattention
Attention is additive in the pair of values
For all :
What the notation means · or point at any symbol
- RReal numbers
- The real numbers: every point on the number line. (
ℝ) - x ∈ TVariables
- s0, s1, v0, v0', v1, v1' range over real numbers: the statement must hold for every choice. (
(s0 s1 v0 v0' v1 v1' : ℝ)) - attend2
- TransformerAttention.attend2: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
TransformerAttention.attend2) - f(x)Function application
- TransformerAttention.attend2 applied to s0, s1, v0 + v0', v1 + v1'. Lean writes f x; papers write f(x). (
TransformerAttention.attend2 s0 s1 (v0 …) - =Equality
- attend2(s0, s1, v0 + v0', v1 + v1') and attend2(s0, s1, v0, v1) + attend2(s0, s1, v0', v1') are the same. (
TransformerAttention.attend2 s0 s1 (v0 …) - +Addition
- v0 plus v0'. (
v0 + v0')
theorem attn2-add-values (s0 s1 v0 v0' v1 v1' : ℝ) :
TransformerAttention.attend2 s0 s1 (v0 + v0') (v1 + v1') = TransformerAttention.attend2 s0 s1 v0 v1 + TransformerAttention.attend2 s0 s1 v0' v1'attend2 of the pointwise sum equals the sum of the attend2 outputs.
How it was proved
- Proved
- 27 September 2026
- By
- Ajax, with the agent “ompmuse”
- Checked by
- Lean's kernel (Lean 4.33.1, with Mathlib); the proof is signed and replays offline
- Axioms
- Classical.choice, Quot.sound, propext
- Proof
- 14 lines of Lean
Sign in to see the proofThe proof, its signed report and its history.