ceos/unsloth
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Update save.py

Browse source
This commit is contained in:
Daniel Han 2024-12-03 16:35:38 -08:00
commit 70893fcd9b
1 changed files with 2 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

3
unsloth/save.py
View file

@ -762,16 +762,17 @@ def install_llama_cpp_make_non_blocking():
# https://github.com/ggerganov/llama.cpp/issues/7062
# Weirdly GPU conversion for GGUF breaks??
# env = { **os.environ, "LLAMA_CUDA": "1", }
n_jobs = max(int(psutil.cpu_count()*1.5), 1)
# Force make clean
check = os.system("make clean -C llama.cpp")
IS_CMAKE = False
if check == 0:
# Uses old MAKE
n_jobs = max(int(psutil.cpu_count()*1.5), 1)
full_command = ["make", "all", "-j"+str(n_jobs), "-C", "llama.cpp"]
IS_CMAKE = False
else:
# Uses new CMAKE
n_jobs = max(int(psutil.cpu_count()), 1) # Use less CPUs since 1.5x faster
check = os.system("cmake llama.cpp -B llama.cpp/build -DBUILD_SHARED_LIBS=OFF -DGGML_CUDA=OFF -DLLAMA_CURL=ON")
if check != 0:
raise RuntimeError(f"*** Unsloth: Failed compiling llama.cpp using os.system(...) with error {check}. Please report this ASAP!")