forked from turboderp/exllama
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtest_benchmark_ppl.sh
executable file
·16 lines (15 loc) · 1.49 KB
/
test_benchmark_ppl.sh
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
echo "-------------------------------------------------------------------------------------------------------------"
python test_benchmark_inference.py -ppl -d /mnt/str/models/llama-7b-4bit-128g
echo "-------------------------------------------------------------------------------------------------------------"
python test_benchmark_inference.py -ppl -d /mnt/str/models/llama-13b-4bit-128g
echo "-------------------------------------------------------------------------------------------------------------"
python test_benchmark_inference.py -ppl -d /mnt/str/models/llama-30b-4bit-128g
echo "-------------------------------------------------------------------------------------------------------------"
python test_benchmark_inference.py -ppl -d /mnt/str/models/llama-30b-4bit-128g-act
echo "-------------------------------------------------------------------------------------------------------------"
python test_benchmark_inference.py -ppl -d /mnt/str/models/llama-30b-4bit-32g-act-ts -l 1550
echo "-------------------------------------------------------------------------------------------------------------"
python test_benchmark_inference.py -ppl -d /mnt/str/models/koala-13B-4bit-128g-act
echo "-------------------------------------------------------------------------------------------------------------"
python test_benchmark_inference.py -ppl -d /mnt/str/models/wizardlm-30b-uncensored-4bit-act-order
echo "-------------------------------------------------------------------------------------------------------------"