Skip to content

A computer, compiled into the weights of a language model. An exact CPU inside Qwen3.5-4B, with Gated DeltaNet state as RAM.

README: Results -> Benchmarks

Divyansh Agrawalcad32e8e2b23
diff --git a/README.md b/README.md
index 48b5195..e27587b 100644
--- a/README.md
+++ b/README.md
@@ -20,7 +20,7 @@ runs the multiplication, and the model writes `478128321` itself, digit by digit
never printed a wrong answer, from 3-digit sums to 20-digit products, and it still loads like any
other Qwen checkpoint.
-## Results
+## Benchmarks
These are Qwen3.5-4B scores, greedy decoding. Each row compares base Qwen and Qwenputer on the
same problems, with a paired McNemar test. The Python-tool column is base Qwen calling a real