mirror of
https://github.com/ikawrakow/ik_llama.cpp.git
synced 2026-02-03 13:04:59 +00:00
154e0d75fc
* Merging mainline - WIP * Merging mainline - WIP AVX2 and CUDA appear to work. CUDA performance seems slightly (~1-2%) lower as it is so often the case with llama.cpp/ggml after some "improvements" have been made. * Merging mainline - fix Metal * Remove check --------- Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
34 lines
1.1 KiB
Markdown
34 lines
1.1 KiB
Markdown
# export-lora
|
|
|
|
Apply LORA adapters to base model and export the resulting model.
|
|
|
|
```
|
|
usage: llama-export-lora [options]
|
|
|
|
options:
|
|
-m, --model model path from which to load base model (default '')
|
|
--lora FNAME path to LoRA adapter (can be repeated to use multiple adapters)
|
|
--lora-scaled FNAME S path to LoRA adapter with user defined scaling S (can be repeated to use multiple adapters)
|
|
-t, --threads N number of threads to use during computation (default: 4)
|
|
-o, --output FNAME output file (default: 'ggml-lora-merged-f16.gguf')
|
|
```
|
|
|
|
For example:
|
|
|
|
```bash
|
|
./bin/llama-export-lora \
|
|
-m open-llama-3b-v2-q8_0.gguf \
|
|
-o open-llama-3b-v2-q8_0-english2tokipona-chat.gguf \
|
|
--lora lora-open-llama-3b-v2-q8_0-english2tokipona-chat-LATEST.gguf
|
|
```
|
|
|
|
Multiple LORA adapters can be applied by passing multiple `--lora FNAME` or `--lora-scaled FNAME S` command line parameters:
|
|
|
|
```bash
|
|
./bin/llama-export-lora \
|
|
-m your_base_model.gguf \
|
|
-o your_merged_model.gguf \
|
|
--lora-scaled lora_task_A.gguf 0.5 \
|
|
--lora-scaled lora_task_B.gguf 0.5
|
|
```
|