Datasets:

erg0dic
/

STaR

Modalities:

Text

Formats:

csv

Size:

1M - 10M

ArXiv:

Libraries:

Datasets

Dask

License:

Dataset card Data Studio Files Files and versions Community

Dataset Viewer

Auto-converted to Parquet

Split (2)

train · 151k rows

Unnamed: 0 int64 0 93.6k	edges stringclasses 9 values	edge_labels stringlengths 10 99	query_edge stringclasses 6 values	query_label stringclasses 21 values
0	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
1	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
2	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
3	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
4	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
5	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
6	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
7	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
8	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
9	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
10	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
11	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
12	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
13	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
14	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
15	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
16	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
17	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
18	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
19	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
20	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
21	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
22	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
23	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
24	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
25	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
26	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
27	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
28	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
29	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
30	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
31	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
32	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
33	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
34	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
35	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
36	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
37	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
38	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
39	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
40	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
41	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
42	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
43	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
44	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
45	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
46	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
47	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
48	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
49	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
50	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
51	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
52	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
53	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
54	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
55	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
56	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
57	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
58	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
59	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
60	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
61	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
62	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
63	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
64	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
65	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
66	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
67	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
68	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
69	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
70	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
71	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
72	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
73	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
74	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
75	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
76	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
77	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
78	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
79	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
80	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
81	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
82	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
83	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
84	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
85	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
86	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
87	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
88	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
89	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
90	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
91	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
92	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
93	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
94	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
95	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
96	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
97	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
98	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=
99	[(0, 1), (1, 2)]	['=', '=']	(0, 2)	=

End of preview. Expand in Data Studio

Systematic, Multipath, disjunctive Spatio-Temporal Reasoning (STaR) benchmark

Main idea

This is part of the paper that was published in ICLR 2025 where further details can be found.

The main idea is to expand the concept of binary relational composition to not just be atomic and require that a model reason over multiple paths instead of a single one in contrast with previous art.

This can be seen from the example below contrasting a popular previous benchmark.

Details

The layout is the same for each dataset in the benchmark. There are two axes of complexity

k: the path length between the source and target node and
b: the number of paths between the source and target nodes

Each dataset has the following rows:

Edge Index	Edge Labels	Query Edge	Query Label
[(0,1), ...]	EQ	(0,5)	TPP

Edge Index: List[Tuple[int, int]]. A list of edges characterizing the graph.
Edge Labels: List[str]. A list of edge labels corresponding to the edge index.
Query Edge: Tuple[int, int]. The target edge that a model needs to predict.
Query Label: str. The label corresponding to the query edge.

The training sets marked by train_*.csv for RCC-8 and Interval algebra (the semigroups that constitute the datasets) contain small graphs of $k=2,3,4$ and $b=1,2,3$.

The rest of the datasets are for testing systematic generalization. The target classes are balanced.

Downloads last month: 63

Size of downloaded dataset files:

2.9 GB

Size of the auto-converted Parquet files:

807 MB

Number of rows:

2,032,800