vrr/prima.cpp
2
0
Fork 0
mirror of https://github.com/Lizonghang/prima.cpp.git synced 2025-09-10 11:44:33 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Enable distributed model perplexity measurement for different bit-width models with -lw and -ngl parameters

Browse source
This commit is contained in:
leeetao 2025-07-01 09:19:19 +00:00
parent 48b7f53abb
commit 82787be7eb
2 changed files with 22 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

4
examples/perplexity/perplexity.cpp
View file

@ -533,8 +533,8 @@ static results_perplexity perplexity(llama_context * ctx, const gpt_params & par
llama_send_meta(ctx, &meta);
LOG_INF("%s: rank 0 tokens_size sent successfully\n", __func__);
} else {
LOG_INF("%s: rank %d waiting 5 seconds for rank 0 to complete tokenization\n", __func__, my_rank);
std::this_thread::sleep_for(std::chrono::milliseconds(5000));
LOG_INF("%s: rank %d waiting 7 seconds for rank 0 to complete tokenization\n", __func__, my_rank);
std::this_thread::sleep_for(std::chrono::milliseconds(7000));
LOG_INF("%s: rank %d delay completed, now receiving tokens_size\n", __func__, my_rank);
if (llama_recv_meta(ctx, &meta) == -1) {
return { {}, -1.0, {}, {} };