Skip to content

Add flag for gpu layers in finetune.cpp #4128

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 1 commit into from
Nov 19, 2023
Merged

Add flag for gpu layers in finetune.cpp #4128

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
1 change: 1 addition & 0 deletions examples/finetune/finetune.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1288,6 +1288,7 @@ static void train_print_usage(int argc, char ** argv, const struct train_params
fprintf(stderr, " --model-base FNAME model path from which to load base model (default '%s')\n", params->fn_model_base);
fprintf(stderr, " --lora-out FNAME path to save llama lora (default '%s')\n", params->fn_lora_out);
fprintf(stderr, " --only-write-lora only save llama lora, don't do any training. use this if you only want to convert a checkpoint to a lora adapter.\n");
fprintf(stderr, " --n-gpu-layers N Number of model layers to offload to GPU (default 0).\n");
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This was put on the https://github.com/ggerganov/llama.cpp/blob/master/common/train.cpp

fprintf(stderr, " --norm-rms-eps F RMS-Norm epsilon value (default %f)\n", params->f_norm_rms_eps);
fprintf(stderr, " --rope-freq-base F Frequency base for ROPE (default %f)\n", params->rope_freq_base);
fprintf(stderr, " --rope-freq-scale F Frequency scale for ROPE (default %f)\n", params->rope_freq_scale);
Expand Down