BitNet b1.58-large TQ2_0

Run (UTC)	geist	llama.cpp	pp512 tok/s	vs llama	tg64 tok/s	vs llama	Files
2026-10-09 18:48	db7f140d5	d81235049	13261.2	-6 %	342.3	-7 %	report · raw
2026-10-09 16:08	53aacaacb	d81235049	6202.9	-56 %	342.0	-7 %	report · raw

Every run of this cell, newest first. pp512: prefill of 512 tokens; tg64: decode of 64 tokens at depth 512; median of 12 samples. The raw file holds every sample and the full host, engine and model provenance. Back to the matrix.