Skip to content

ggml : set output of view src #25729

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
ggerganov merged 9 commits into from
Jul 28, 2026
Merged

ggml : set output of view src #25729

Changes from 2 commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
2ee90a9
llama-graph: set_outputs to t->view_src
am17an Jul 15, 2026
fd4fd67
change set_output to GGML_ASSERT about views not being outputs
am17an Jul 16, 2026
3c430ec
Merge branch 'master' into pr/25729
ggerganov Jul 28, 2026
28a62f9
sampler : avoid views in outputs
ggerganov Jul 28, 2026
f3aa737
cont : fix dist sampler
ggerganov Jul 28, 2026
8fc77ee
cont : consistent logits handling
ggerganov Jul 28, 2026
299077b
ggml : set output of view src
gaugarg-nv Jul 28, 2026
bf17c7d
graph : simplify set_outputs()
ggerganov Jul 28, 2026
7255f0a
cont : cleanup
ggerganov Jul 28, 2026
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
23 changes: 14 additions & 9 deletions src/llama-graph.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1228,45 +1228,50 @@ void llm_graph_result::set_inputs(const llama_ubatch * ubatch) {
}

void llm_graph_result::set_outputs(const llm_graph_params & params) {
auto set_output = [](ggml_tensor * t) {
ggml_set_output(t);
GGML_ASSERT(t->view_src == nullptr && "views cannot be outputs");
};

if (t_logits != nullptr) {
ggml_set_output(t_logits);
set_output(t_logits);
}
if (t_embd != nullptr) {
ggml_set_output(t_embd);
set_output(t_embd);
}
if (t_embd_pooled != nullptr) {
ggml_set_output(t_embd_pooled);
set_output(t_embd_pooled);
}
if (t_h_nextn != nullptr) {
ggml_set_output(t_h_nextn);
set_output(t_h_nextn);
}
{
const auto & embeddings_layer_inp = params.cparams.embeddings_layer_inp;
for (size_t il = 0; il < embeddings_layer_inp.size(); ++il) {
if (embeddings_layer_inp[il]) {
GGML_ASSERT(t_layer_inp[il] != nullptr && "layer input tensor is null");
ggml_set_output(t_layer_inp[il]);
set_output(t_layer_inp[il]);
}
}
}
for (auto & [seq_id, t] : t_sampled) {
if (t != nullptr) {
ggml_set_output(t);
set_output(t);
}
}
for (auto & [seq_id, t] : t_sampled_probs) {
if (t != nullptr) {
ggml_set_output(t);
set_output(t);
}
}
for (auto & [seq_id, t] : t_sampled_logits) {
if (t != nullptr) {
ggml_set_output(t);
set_output(t);
}
}
for (auto & [seq_id, t] : t_candidates) {
if (t != nullptr) {
ggml_set_output(t);
set_output(t);
}
}
}
Expand Down
Loading