-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathresults_specinfer_greedy.log
36 lines (36 loc) · 4.51 KB
/
results_specinfer_greedy.log
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
[2024-02-10 17:27:25,588] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to cuda (auto detect)
Namespace(model='JackFram/llama-68m', target='meta-llama/Llama-2-7b-hf', dataset='dataset/c4_small.json', growmap='/home/zhuominc/workspace/Sequoia/growmaps/5x8-tree.pt', start=0, end=200, T=0.6, P=1.0, DP=0.99, D=1, B=10, W=32, M=384, Mode='greedy', decay=0.85, negative=False, static=False, offloading=False)
41
total time :165.01635s, latency :0.00744s, decoding step: 22190, large model step: 5617
[2024-02-10 17:31:22,435] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to cuda (auto detect)
Namespace(model='JackFram/llama-68m', target='meta-llama/Llama-2-7b-hf', dataset='cnn', growmap='/home/zhuominc/workspace/Sequoia/growmaps/5x8-tree.pt', start=0, end=200, T=0.6, P=1.0, DP=0.99, D=1, B=10, W=32, M=384, Mode='greedy', decay=0.85, negative=False, static=False, offloading=False)
41
total time :252.47040s, latency :0.00968s, decoding step: 26070, large model step: 8635
[2024-02-10 17:45:05,565] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to cuda (auto detect)
Namespace(model='JackFram/llama-68m', target='meta-llama/Llama-2-7b-hf', dataset='openwebtext', growmap='/home/zhuominc/workspace/Sequoia/growmaps/5x8-tree.pt', start=0, end=200, T=0.6, P=1.0, DP=0.99, D=1, B=10, W=32, M=384, Mode='greedy', decay=0.85, negative=False, static=False, offloading=False)
41
total time :236.96769s, latency :0.00922s, decoding step: 25697, large model step: 8075
[2024-02-10 17:49:33,251] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to cuda (auto detect)
Namespace(model='JackFram/llama-68m', target='meta-llama/Llama-2-13b-hf', dataset='dataset/c4_small.json', growmap='/home/zhuominc/workspace/Sequoia/growmaps/5x8-tree.pt', start=0, end=200, T=0.6, P=1.0, DP=0.99, D=1, B=10, W=32, M=384, Mode='greedy', decay=0.85, negative=False, static=False, offloading=False)
41
total time :299.58937s, latency :0.01380s, decoding step: 21710, large model step: 5770
[2024-02-10 17:54:58,415] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to cuda (auto detect)
Namespace(model='JackFram/llama-68m', target='meta-llama/Llama-2-13b-hf', dataset='cnn', growmap='/home/zhuominc/workspace/Sequoia/growmaps/5x8-tree.pt', start=0, end=200, T=0.6, P=1.0, DP=0.99, D=1, B=10, W=32, M=384, Mode='greedy', decay=0.85, negative=False, static=False, offloading=False)
41
total time :449.33620s, latency :0.01728s, decoding step: 26002, large model step: 8669
[2024-02-10 18:02:56,041] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to cuda (auto detect)
Namespace(model='JackFram/llama-68m', target='meta-llama/Llama-2-13b-hf', dataset='openwebtext', growmap='/home/zhuominc/workspace/Sequoia/growmaps/5x8-tree.pt', start=0, end=200, T=0.6, P=1.0, DP=0.99, D=1, B=10, W=32, M=384, Mode='greedy', decay=0.85, negative=False, static=False, offloading=False)
41
total time :444.32544s, latency :0.01733s, decoding step: 25642, large model step: 8571
[2024-02-10 18:10:46,081] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to cuda (auto detect)
Namespace(model='JackFram/llama-160m', target='meta-llama/Llama-2-13b-hf', dataset='dataset/c4_small.json', growmap='/home/zhuominc/workspace/Sequoia/growmaps/5x8-tree.pt', start=0, end=200, T=0.6, P=1.0, DP=0.99, D=1, B=10, W=32, M=384, Mode='greedy', decay=0.85, negative=False, static=False, offloading=False)
41
total time :313.23764s, latency :0.01439s, decoding step: 21768, large model step: 4927
[2024-02-10 18:16:30,170] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to cuda (auto detect)
Namespace(model='JackFram/llama-160m', target='meta-llama/Llama-2-13b-hf', dataset='cnn', growmap='/home/zhuominc/workspace/Sequoia/growmaps/5x8-tree.pt', start=0, end=200, T=0.6, P=1.0, DP=0.99, D=1, B=10, W=32, M=384, Mode='greedy', decay=0.85, negative=False, static=False, offloading=False)
41
total time :462.56431s, latency :0.01775s, decoding step: 26057, large model step: 7293
[2024-02-10 18:24:45,638] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to cuda (auto detect)
Namespace(model='JackFram/llama-160m', target='meta-llama/Llama-2-13b-hf', dataset='openwebtext', growmap='/home/zhuominc/workspace/Sequoia/growmaps/5x8-tree.pt', start=0, end=200, T=0.6, P=1.0, DP=0.99, D=1, B=10, W=32, M=384, Mode='greedy', decay=0.85, negative=False, static=False, offloading=False)
41
total time :463.36217s, latency :0.01805s, decoding step: 25676, large model step: 7306