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-Chen 2024-03-16 23:17:35 +11:00
commit 7e9f092e9f
1 changed files with 4 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

7
unsloth/save.py
View file

@ -730,7 +730,10 @@ def save_to_gguf(
install_llama_cpp_blocking()
pass
# Check if successful. If not install 10th latest release
if error != 0 or not os.path.exists("llama.cpp/quantize"): install_llama_cpp_old(-10)
if error != 0 or not os.path.exists("llama.cpp/quantize"):
print(f"Unsloth: llama.cpp error code = {error}.")
install_llama_cpp_old(-10)
pass
if quantization_method == "f32": first_conversion = "f32"
elif quantization_method == "f16": first_conversion = "f16"
@ -768,9 +771,7 @@ def save_to_gguf(
f"--outtype {first_conversion} --concurrency {n_cpus}"
else:
# Need to fix convert-hf-to-gguf.py for some models!
print("Fixing Gemma")
_fix_gemma_gguf()
print("Fixed!")
command = f"python llama.cpp/convert-hf-to-gguf.py {model_directory} "\
f"--outfile {final_location} "\