wmantly/llama.cpp-xrip
1
0 Fork 0
Code Issues Pull Requests Actions 30 Packages Projects Releases Wiki Activity

sampler: reduce the size of the probe (#29285)

Browse Source
This commit is contained in:
Max Krasnyansky authored and GitHub committed 2026-09-23 08:12:04 +03:00
1 parent e6ab7c1a41
commit 441df11f65
1 file changed
+2 -2
Download Patch File Download Diff File Expand all files Collapse all files
src/llama-sampler.cpp
+2 -2
View File
@@ -644,7 +644,7 @@ static bool llama_sampler_backend_support(
return true;
}
auto probe = llama_sampler_backend_probe_graph(smpl, 1024*1024, GGML_DEFAULT_GRAPH_SIZE, true);
auto probe = llama_sampler_backend_probe_graph(smpl, 128*1024, GGML_DEFAULT_GRAPH_SIZE, true);
for (int i = 0; i < ggml_graph_n_nodes(probe.gf); i++) {
struct ggml_tensor * op = ggml_graph_node(probe.gf, i);
@@ -764,7 +764,7 @@ static bool llama_sampler_chain_backend_init(
res = res && cur_prefix;
}
auto probe = llama_sampler_backend_probe_graph(smpl, 1024*1024, GGML_DEFAULT_GRAPH_SIZE, false);
auto probe = llama_sampler_backend_probe_graph(smpl, 128*1024, GGML_DEFAULT_GRAPH_SIZE, false);
chain->n_nodes = llama_sampler_backend_probe_n_nodes(probe);
return res;