M1 (7-core GPU, 8 GB)

2 records across 2 models (quants: Q4_K - Medium).

hardwaremodelquantbackendtok/spp tok/sWttft sctx/tgdateclassflagssource
M1 (7-core GPU, 8 GB)llama-3-1-8b-instruct Q4_K - Mediumllamafile13.385236666666666108.71905633333331–11.513906754444445––sourced–source
M1 (7-core GPU, 8 GB)llama-3-2-1b-instruct Q4_K - Mediumllamafile38.51773918518518532.9676827037036–2.4294918625925925––sourced–source

All rows cite their source. Provenance: community 40018 sourced 1085

← all hardware