Skip to content

batch : fix check for empty sequences in memory #14364

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 2 commits into from
Jun 24, 2025
Merged

batch : fix check for empty sequences in memory #14364

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
7d7003d
batch : fix check for empty sequences in memory
ggerganov Jun 24, 2025
53608db
cont : reuse the var
ggerganov Jun 24, 2025
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
10 changes: 6 additions & 4 deletions src/llama-batch.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -244,19 +244,21 @@ bool llama_batch_allocr::init(
continue;
}

if (memory) {
const llama_pos p0 = memory ? memory->seq_pos_max(s) : -1;

if (p0 >= 0) {
bool ok = true;

if (batch.token) {
if (seq_pos_min(s) != memory->seq_pos_max(s) + 1) {
if (seq_pos_min(s) != p0 + 1) {
ok = false;
}
} else {
assert(batch.embd);

// for embeddings (typically used as vision input), we allow them to have repeating positions
// ref: https://github.com/ggml-org/llama.cpp/issues/13694#issuecomment-2983871762
if (seq_pos_min(s) != memory->seq_pos_max(s) && seq_pos_min(s) != memory->seq_pos_max(s) + 1) {
if (seq_pos_min(s) != p0 && seq_pos_min(s) != p0 + 1) {
ok = false;
}
}
Expand All @@ -267,7 +269,7 @@ bool llama_batch_allocr::init(
" - the last position stored in the memory module of the context (i.e. the KV cache) for sequence %d is X = %d\n"
" - the tokens for sequence %d in the input batch have a starting position of Y = %d\n"
" it is required that the sequence positions remain consecutive: Y = X + 1\n",
__func__, s, s, memory->seq_pos_max(s), s, seq_pos_min(s));
__func__, s, s, p0, s, seq_pos_min(s));

return false;
}
Expand Down
Loading