Skip to content

dflash: fix VL/multimodal crash by clearing draft on position gap #7

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

Closed
dblmca wants to merge 1 commit into from
+22 −0
Closed

dflash: fix VL/multimodal crash by clearing draft on position gap #7

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
22 changes: 22 additions & 0 deletions common/speculative.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1031,6 +1031,28 @@ struct common_speculative_impl_draft_dflash : public common_speculative_impl {

const int32_t n_ubatch = (int32_t) llama_n_ubatch(ctx_dft);

// VL support: handle position gaps from image/audio embeddings.
// Embedding batches are skipped (line ~1008), so the draft KV cache
// misses positions filled by image tokens in the target. Detect any
// non-consecutive position and clear the draft sequence so processing
// can rebuild from the current text position.
for (llama_seq_id seq_id = 0; seq_id < (llama_seq_id) n_seq; ++seq_id) {
if (i_batch_beg[seq_id] < 0) {
continue;
}
GGML_ASSERT(batch_in.embd == nullptr);
const llama_pos dft_pos_max = llama_memory_seq_pos_max(
llama_get_memory(ctx_dft), seq_id);
const llama_pos batch_pos = batch_in.pos[i_batch_beg[seq_id]];
if (dft_pos_max >= 0 && batch_pos != dft_pos_max + 1) {
LOG_DBG("%s: non-consecutive pos for seq %d "
"(dft_pos_max=%d, batch_pos=%d) — clearing draft\n",
__func__, (int) seq_id,
(int) dft_pos_max, (int) batch_pos);
llama_memory_seq_rm(llama_get_memory(ctx_dft), seq_id, -1, -1);
}
}

for (llama_seq_id seq_id = 0; seq_id < (llama_seq_id) n_seq; ++seq_id) {
if (i_batch_beg[seq_id] < 0) {
continue;
Expand Down