llama.cpp/examples/perplexity/perplexity.cpp at ik/faster_hellaswag

mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-08-01 23:50:47 -05:00

Files

Iwan Kawrakow ccc78a200e hellaswag: speed up even more by parallelizing log-prob evaluation

For Mistral-7B and fp16, time on my system goes down from 536 seconds
to 423 seconds for the full evaluation dataset (10042 tasks).

2024-01-18 18:25:29 +02:00

View Raw