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

Update tools.sh

Browse source 
Missing conversion part from llama model into ggml. 
Added with validation in case it exist to skip to quantization.
This commit is contained in:
RSereno 2023-03-25 16:01:11 +00:00 • committed by GitHub
parent 4eae17153c
commit 8daa71d958
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 10 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

11
.devops/tools.sh
View file

@ -10,6 +10,8 @@ shift
# Join the remaining arguments into a single string # Join the remaining arguments into a single string
arg2="$@" arg2="$@"
modelname="ggml-model-f16.bin"
if [[ $arg1 == '--convert' || $arg1 == '-c' ]]; then if [[ $arg1 == '--convert' || $arg1 == '-c' ]]; then
python3 ./convert-pth-to-ggml.py $arg2 python3 ./convert-pth-to-ggml.py $arg2
elif [[ $arg1 == '--quantize' || $arg1 == '-q' ]]; then elif [[ $arg1 == '--quantize' || $arg1 == '-q' ]]; then
@ -18,7 +20,14 @@ elif [[ $arg1 == '--run' || $arg1 == '-r' ]]; then
./main $arg2 ./main $arg2
elif [[ $arg1 == '--all-in-one' || $arg1 == '-a' ]]; then elif [[ $arg1 == '--all-in-one' || $arg1 == '-a' ]]; then
echo "Converting PTH to GGML..." echo "Converting PTH to GGML..."
for i in `ls $1/$2/consolidated.*.pth`; do if [ -f "$1/$2/$modelname" ]; then
echo "Skip model conversion, it already exists: ${i/f16/q4_0}"
else
echo "Converting PTH to GGML: $i into ${i/f16/q4_0}..."
python3 ./convert-pth-to-ggml.py $1/$2 1
fi
echo "Quantizing f16 to q4_0..."
for i in `ls $1/$2/$modelname*`; do
if [ -f "${i/f16/q4_0}" ]; then if [ -f "${i/f16/q4_0}" ]; then
echo "Skip model quantization, it already exists: ${i/f16/q4_0}" echo "Skip model quantization, it already exists: ${i/f16/q4_0}"
else else