mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2025-08-20 06:36:48 -04:00
convert : auto-determine model name based on dir + scripts update
This commit is contained in:
31
scripts/run-all-perf.sh
Executable file
31
scripts/run-all-perf.sh
Executable file
@@ -0,0 +1,31 @@
|
||||
#!/bin/bash
|
||||
|
||||
qnt=(f16 q8_0 q6_k q5_k q5_1 q5_0 q4_k q4_1 q4_0 q3_k q2_k)
|
||||
args="-ngl 999 -n 64 -p 512"
|
||||
|
||||
if [ -z "$1" ]; then
|
||||
echo "usage: $0 <model> [qnt] [args]"
|
||||
echo "default: $0 <model> \"${qnt[@]}\" \"${args}\""
|
||||
exit 1
|
||||
fi
|
||||
|
||||
if [ ! -z "$2" ]; then
|
||||
qnt=($2)
|
||||
fi
|
||||
|
||||
if [ ! -z "$3" ]; then
|
||||
args="$3"
|
||||
fi
|
||||
|
||||
model="$1"
|
||||
out="../tmp/results-${model}"
|
||||
|
||||
mkdir -p ${out}
|
||||
|
||||
mstr=""
|
||||
|
||||
for q in ${qnt[@]}; do
|
||||
mstr="${mstr} -m ../models/${model}/ggml-model-${q}.gguf"
|
||||
done
|
||||
|
||||
./bin/llama-bench ${mstr} ${args} 2> /dev/null
|
Reference in New Issue
Block a user