Skip to content

context : always use non-causal attention for encoder graphs #12447

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
Mar 18, 2025
Merged

context : always use non-causal attention for encoder graphs #12447

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
a0554c3
context : always use non-causal attention for encoder graphs
ggerganov Mar 18, 2025
29acf2c
context : move the change to llama_context::encode()
ggerganov Mar 18, 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
9 changes: 9 additions & 0 deletions src/llama-context.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1057,13 +1057,22 @@ int llama_context::encode(llama_batch & inp_batch) {
ggml_backend_sched_reset(sched.get());
ggml_backend_sched_set_eval_callback(sched.get(), cparams.cb_eval, cparams.cb_eval_user_data);

const auto causal_attn_org = cparams.causal_attn;

// always use non-causal attention for encoder graphs
// TODO: this is a tmp solution until we have a proper way to support enc-dec models
// ref: https://github.com/ggml-org/llama.cpp/pull/12181#issuecomment-2730451223
cparams.causal_attn = false;

auto * gf = graph_init();
auto res = graph_build(ctx_compute.get(), gf, ubatch, LLM_GRAPH_TYPE_ENCODER);

ggml_backend_sched_alloc_graph(sched.get(), gf);

res->set_inputs(&ubatch);

cparams.causal_attn = causal_attn_org;

const auto compute_status = graph_compute(gf, n_tokens > 1);
switch (compute_status) {
case GGML_STATUS_SUCCESS:
Expand Down
Loading