Skip to content

Distribute and run LLMs with a single file.

Disable warmup

Justine Tunney21a30bed3dbc
diff --git a/llama.cpp/common.h b/llama.cpp/common.h
index 465e1b8c..526091da 100644
--- a/llama.cpp/common.h
+++ b/llama.cpp/common.h
@@ -171,7 +171,7 @@ struct gpt_params {
bool infill = false; // use infill mode
bool dump_kv_cache = false; // dump the KV cache contents for debugging purposes
bool no_kv_offload = false; // disable KV offloading
- bool warmup = true; // warmup run
+ bool warmup = false; // warmup run
bool check_tensors = false; // validate tensor data
std::string cache_type_k = "f16"; // KV cache data type for the K
diff --git a/llamafile/schlep.c b/llamafile/schlep.c
index 79358379..41f1bf3c 100644
--- a/llamafile/schlep.c
+++ b/llamafile/schlep.c
@@ -80,6 +80,10 @@ void llamafile_schlep(const void *data, size_t size) {
if (size < 128 * 1024 * 1024)
return;
+ // don't bother if we're using a gpu
+ if (llamafile_has_gpu())
+ return;
+
// don't bother if stderr isn't a terminal
if (!isatty(2))
return;