vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

llama : enable GPU inference by default with Metal

Browse source
This commit is contained in:
Georgi Gerganov 2023-09-03 10:30:53 +03:00
parent 15f1790a75
commit 99161230c4
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
5 changed files with 21 additions and 18 deletions
Download patch file Download diff file Expand all files Collapse all files

15
examples/main/main.cpp
View file

@ -151,14 +151,6 @@ int main(int argc, char ** argv) {
LOG_TEE("%s: warning: scaling RoPE frequency by %g (default 1.0)\n", __func__, params.rope_freq_scale);
}
if (params.n_ctx > 2048) {
// TODO: determine the actual max context of the model (e.g. 4096 for LLaMA v2) and use that instead of 2048
LOG_TEE("%s: warning: base model only supports context sizes no greater than 2048 tokens (%d specified)\n", __func__, params.n_ctx);
} else if (params.n_ctx < 8) {
LOG_TEE("%s: warning: minimum context size is 8, using minimum size.\n", __func__);
params.n_ctx = 8;
}
LOG_TEE("%s: build = %d (%s)\n", __func__, BUILD_NUMBER, BUILD_COMMIT);
if (params.seed == LLAMA_DEFAULT_SEED) {
@ -194,6 +186,13 @@ int main(int argc, char ** argv) {
return 1;
}
if (params.n_ctx > llama_n_ctx(ctx)) {
LOG_TEE("%s: warning: base model only supports context sizes no greater than %d tokens (%d specified)\n", __func__, llama_n_ctx(ctx), params.n_ctx);
} else if (params.n_ctx < 8) {
LOG_TEE("%s: warning: minimum context size is 8, using minimum size.\n", __func__);
params.n_ctx = 8;
}
// print system information
{
LOG_TEE("\n");