ABCDEFGHIJKLMNOPQRSTUVWXYZ
1
Validation set
2
Total size (prompts)1,500
3
Total size (samples)525,000
4
Number of models3
5
Unique models
meta-llama/Meta-Llama-3-70B-Instruct, meta-llama/Meta-Llama-3-8B-Instruct, mistralai/Mistral-7B-Instruct-v0.3
6
Number of datasets5
7
Unique datasets
wikipedia_de, wikipedia_en, wikipedia_es, wikipedia_fr, wikipedia_ru
8
Number of alternatives5
9
Unique alternatives
fp16, int8, nf4, fp32, watermark
10
% with logprobs67
11
12
13
Test set
14
Total size (prompts)23,200
15
Total size (samples)1,110,000
16
Number of models5
17
Unique models
meta-llama/Meta-Llama-3-70B-Instruct, meta-llama/Meta-Llama-3-8B-Instruct, meta-llama/Meta-Llama-3.1-70B-Instruct, meta-llama/Meta-Llama-3.1-8B-Instruct, mistralai/Mistral-7B-Instruct-v0.3
18
Number of datasets7
19
Unique datasets
wikipedia_de, wikipedia_en, wikipedia_es, wikipedia_fr, wikipedia_ru, humaneval, ultrachat
20
Number of alternatives13
21
Unique alternatives
amazon, azure, deepinfra, fireworks, fp16, groq, int8, nf4, fp32, perplexity, replicate, together, watermark
22
% with logprobs0
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