-
Notifications
You must be signed in to change notification settings - Fork 0
/
route.sh
61 lines (54 loc) · 1.39 KB
/
route.sh
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
dataset=spider
flags=1_1_1_1 # 0_0_0_0 for TS, 1_1_0_0 for SL, 0_0_1_0 for NC, 0_0_0_1 for CW
for dataset in spider bird
do
for flags in 0_0_0_0 1_1_1_1
do
data_path=your data_path
mode=dev
output_path=your output_path
gpus=1
batch_size=32
LLM_model=your sft model path
tag=llama3_8b_MSFT # if online in tag, customize your online LLM
# for sft/msft LLM
CUDA_VISIBLE_DEVICES=0 python route.py --data_path $data_path \
--dataset $dataset \
--output_path $output_path \
--mode $mode \
--LLM_model $LLM_model \
--batch_size $batch_size\
--tag $tag \
--gpus $gpus\
--flags $flags \
--eval_sft 1\
batch_size=32
LLM_model=your instruct model path
tag=llama3_8b_instruct
# for LLM w/o sft/msft
CUDA_VISIBLE_DEVICES=0 python route.py --data_path $data_path \
--dataset $dataset \
--output_path $output_path \
--mode $mode \
--LLM_model $LLM_model \
--batch_size $batch_size\
--tag $tag \
--gpus $gpus\
--flags $flags \
--eval_sft 0\
batch_size=16
LLM_model=qwen2.5-14b-msft
tag=qwen2.5_14b_MSFT # if online in tag, customize your online LLM
# for sft/msft LLM
CUDA_VISIBLE_DEVICES=0 python route.py --data_path $data_path \
--dataset $dataset \
--output_path $output_path \
--mode $mode \
--LLM_model $LLM_model \
--batch_size $batch_size\
--tag $tag \
--gpus $gpus\
--flags $flags \
--eval_sft 1\
done
done