From 4270136948be36221baf36499b8fb1a4a5b7a923 Mon Sep 17 00:00:00 2001 From: Davide Cappellini Date: Fri, 25 Sep 2026 08:43:48 +0200 Subject: [PATCH] docs: correct counted-SBC decay amortised cost figure --- docs/bitbrain_counted_sbc.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/docs/bitbrain_counted_sbc.md b/docs/bitbrain_counted_sbc.md index 6e09244..4aa1536 100644 --- a/docs/bitbrain_counted_sbc.md +++ b/docs/bitbrain_counted_sbc.md @@ -61,7 +61,7 @@ alternatives for three reasons: The bitset `infer` is unchanged. **Config (which is which).** The default is and remains `smBitset`; the bitset -path is byte-for-byte unchanged. Counted mode is selected explicitly or at +path's behaviour is exactly unchanged. Counted mode is selected explicitly or at runtime: | knob | kind | values | @@ -92,8 +92,8 @@ The gun-sized byte-per-cell cost is 12 MiB — acceptable for a gun (and the whole 12.1 MiB model fits comfortably next to the rest of a bot). The MNIST 240 MiB is fine for an offline measurement machine but is 8× the bit tensor; packed 4-bit (INFERRED, not implemented) would halve both figures. At -MNIST scale, `decayEvery = 2000` costs ≈ 0.2 ms/learn amortised (measured train -30.3 s vs 25.4 s over 60k for the decay arm, i.e. ~0.08 ms/sample extra). +MNIST scale, `decayEvery = 2000` costs ≈ 0.07–0.08 ms/learn amortised (measured +train 28–30 s vs 24–25 s over 60k for the decay arm). ---