forked from EleutherAI/lm-evaluation-harness
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathlaunch_rand_smooth_llama2_alpaca_smooth5.sh
106 lines (90 loc) · 4.54 KB
/
launch_rand_smooth_llama2_alpaca_smooth5.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
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
##### base run
python main.py \
--no_cache \
--device=cuda:0 \
--model=hf-causal \
--model_args=pretrained=/fs/cml-projects/broken-token/lm-defenses/llama_cml/trained_models_hf/llama2_Alpaca_7b_baseline_updated_HF,trust_remote_code=True \
--tasks=arc_challenge \
--num_fewshot=25 \
--batch_size=2 \
--output_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_baseline_updated_HF_arc_challenge.json \
--write_out \
--output_base_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_baseline_updated_HF_arc_challenge
python main.py \
--no_cache \
--device=cuda:0 \
--model=hf-causal \
--model_args=pretrained=/fs/cml-projects/broken-token/lm-defenses/llama_cml/trained_models_hf/llama2_Alpaca_7b_baseline_updated_HF,trust_remote_code=True \
--tasks=hellaswag \
--num_fewshot=10 \
--batch_size=2 \
--output_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_baseline_updated_HF_hellaswag.json \
--write_out \
--output_base_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_baseline_updated_HF_hellaswag
python main.py \
--no_cache \
--device=cuda:0 \
--model=hf-causal \
--model_args=pretrained=/fs/cml-projects/broken-token/lm-defenses/llama_cml/trained_models_hf/llama2_Alpaca_7b_baseline_updated_HF,trust_remote_code=True \
--tasks=hendrycksTest-* \
--num_fewshot=5 \
--batch_size=2 \
--output_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_baseline_updated_HF_mmlu.json \
--write_out \
--output_base_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_baseline_updated_HF_mmlu
python main.py \
--no_cache \
--device=cuda:0 \
--model=hf-causal \
--model_args=pretrained=/fs/cml-projects/broken-token/lm-defenses/llama_cml/trained_models_hf/llama2_Alpaca_7b_baseline_updated_HF,trust_remote_code=True \
--tasks=truthfulqa_mc \
--num_fewshot=0 \
--batch_size=2 \
--output_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_baseline_updated_HF_truthfulqa.json \
--write_out \
--output_base_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_baseline_updated_HF_truthfulqa
#### with rand smoothing
python main.py \
--no_cache \
--device=cuda:0 \
--model=hf-causal \
--model_args=pretrained=/fs/cml-projects/broken-token/lm-defenses/llama_cml/trained_models_hf/llama2_Alpaca_7b_noise5_updated_HF,trust_remote_code=True \
--tasks=arc_challenge \
--num_fewshot=25 \
--batch_size=2 \
--output_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_noise5_updated_HF_arc_challenge.json \
--write_out \
--output_base_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_noise5_updated_HF_arc_challenge
python main.py \
--no_cache \
--device=cuda:0 \
--model=hf-causal \
--model_args=pretrained=/fs/cml-projects/broken-token/lm-defenses/llama_cml/trained_models_hf/llama2_Alpaca_7b_noise5_updated_HF,trust_remote_code=True \
--tasks=hellaswag \
--num_fewshot=10 \
--batch_size=2 \
--output_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_noise5_updated_HF_hellaswag.json \
--write_out \
--output_base_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_noise5_updated_HF_hellaswag
python main.py \
--no_cache \
--device=cuda:0 \
--model=hf-causal \
--model_args=pretrained=/fs/cml-projects/broken-token/lm-defenses/llama_cml/trained_models_hf/llama2_Alpaca_7b_noise5_updated_HF,trust_remote_code=True \
--tasks=hendrycksTest-* \
--num_fewshot=5 \
--batch_size=2 \
--output_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_noise5_updated_HF_mmlu.json \
--write_out \
--output_base_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_noise5_updated_HF_mmlu
python main.py \
--no_cache \
--device=cuda:0 \
--model=hf-causal \
--model_args=pretrained=/fs/cml-projects/broken-token/lm-defenses/llama_cml/trained_models_hf/llama2_Alpaca_7b_noise5_updated_HF,trust_remote_code=True \
--tasks=truthfulqa_mc \
--num_fewshot=0 \
--batch_size=2 \
--output_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_noise5_updated_HF_truthfulqa.json \
--write_out \
--output_base_path=/cmlscratch/jkirchen/adv-llm-root/output/rand_smooth/llama2_Alpaca_7b_noise5_updated_HF_truthfulqa