2 ms·
According to few papers and https://github.com/ggerganov/llama.cpp/pull/1684 https://github.com/ggerganov/llama.cpp/pull/1684, ~2.5GB 7B parameters model size h
by YetAnotherNick 3y ago
According to few papers and https://github.com/ggerganov/llama.cpp/pull/1684 https://github.com/ggerganov/llama.cpp/pull/1684, ~2.5GB 7B parameters model size has the same performance as baseline 7B model(with 14GB size).