Skip to content

Temperature: Fix Needlessly Iterating on Candidates During Greedy Sampling #1854

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Nov 22, 2025
Merged

Temperature: Fix Needlessly Iterating on Candidates During Greedy Sampling #1854

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 2 additions & 10 deletions gpttype_adapter.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1540,13 +1540,10 @@ void sample_entropy(llama_token_data_array * cur_p, float min_temp, float max_te

void sample_temperature(llama_token_data_array * candidates_p, float temp, float smoothing_factor, float smoothing_curve)
{
bool isgreedy = false;
if (temp <= 0)
{
// Imitate greedy sampling
temp = 0.00390625f; //cannot be zero else div0, this is 1/256
smoothing_factor = 0;
isgreedy = true;
sample_top_k(candidates_p, 1); //only want first candidate
return;
}

for (size_t i = 0; i < candidates_p->size; ++i) {
Expand All @@ -1565,11 +1562,6 @@ void sample_temperature(llama_token_data_array * candidates_p, float temp, float
}
sample_softmax(candidates_p);
}

if(isgreedy)
{
sample_top_k(candidates_p, 1); //only want first candidate
}
}

void sample_grammar(FileFormat file_format, int32_t n_vocab, llama_token_data_array * candidates, const struct llama_grammar * grammar) {
Expand Down