neoneye commited on
Commit
96cde5f
1 Parent(s): d13c6c2

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +179 -3
README.md CHANGED
@@ -1,3 +1,179 @@
1
- ---
2
- license: mit
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ task_categories:
4
+ - image-to-text
5
+ - text-to-image
6
+ language:
7
+ - en
8
+ pretty_name: simons ARC (abstraction & reasoning corpus) object mass version 23
9
+ size_categories:
10
+ - 10K<n<100K
11
+ configs:
12
+ - config_name: default
13
+ data_files:
14
+ - split: train
15
+ path: data.jsonl
16
+ ---
17
+
18
+ # Version 1
19
+
20
+ Measure the mass of objects for `pixel connectivity 4` and `pixel connectivity 8`.
21
+
22
+ image size: 1-10.
23
+
24
+ max_mass: 4.
25
+
26
+ # Version 2
27
+
28
+ image size: 1-20.
29
+
30
+ max_mass: 5.
31
+
32
+ This converged too slowly. I was too optimistic. I will have to proceed slower.
33
+
34
+ # Version 3
35
+
36
+ image size: 1-12.
37
+
38
+ max_mass: 5.
39
+
40
+ Too big spikes in the training loss. I will have to lower the `max_mass`, and gradually increase it.
41
+
42
+ # Version 4
43
+
44
+ image size: 1-15.
45
+
46
+ max_mass: 2.
47
+
48
+ The validation loss for this is fine. I'm increasing the `max_mass`.
49
+
50
+ # Version 5
51
+
52
+ image size: 1-20.
53
+
54
+ max_mass: 3.
55
+
56
+ # Version 6
57
+
58
+ image size: 1-25.
59
+
60
+ max_mass: 4.
61
+
62
+ # Version 7
63
+
64
+ image size: 1-30.
65
+
66
+ max_mass: 2.
67
+
68
+ # Version 8
69
+
70
+ image size: 1-30.
71
+
72
+ max_mass: 7.
73
+
74
+ # Version 9
75
+
76
+ image size: 1-30.
77
+
78
+ max_mass: 8.
79
+
80
+ # Version 10
81
+
82
+ image size: 1-30.
83
+
84
+ max_mass: 9.
85
+
86
+ # Version 11
87
+
88
+ image size: 1-30.
89
+
90
+ max_mass: 10.
91
+
92
+ # Version 12
93
+
94
+ image size: 1-30.
95
+
96
+ max_mass: 11.
97
+
98
+ # Version 13
99
+
100
+ image size: 1-30.
101
+
102
+ max_mass: 15.
103
+
104
+ # Version 14
105
+
106
+ image size: 1-30.
107
+
108
+ max_mass: 20.
109
+
110
+ # Version 15
111
+
112
+ image size: 1-30.
113
+
114
+ max_mass: 25.
115
+
116
+ # Version 16
117
+
118
+ Added `PixelConnectivity.CORNER4`, so it can identify diagonal structures.
119
+
120
+ image size: 1-10.
121
+
122
+ max_mass: 5.
123
+
124
+ # Version 17
125
+
126
+ image size: 1-15.
127
+
128
+ max_mass: 10.
129
+
130
+ # Version 18
131
+
132
+ Added mulple new `PixelConnectivity` modes: `LR2`, `TB2`, `TLBR2`, `TRBL2`, so it can identify line structures.
133
+
134
+ image size: 1-10.
135
+
136
+ max_mass: 5.
137
+
138
+ # Version 19
139
+
140
+ image size: 1-15.
141
+
142
+ max_mass: 10.
143
+
144
+ # Version 20
145
+
146
+ Disabled `NEAREST4`, `ALL8`, `CORNER4`. So the model can be trained only on line structures.
147
+
148
+ image size: 1-20.
149
+
150
+ max_mass: 15.
151
+
152
+ The model struggles with these settings. I have to lower the settings.
153
+
154
+ # Version 21
155
+
156
+ Still with disabled `NEAREST4`, `ALL8`, `CORNER4`. So the model can be trained only on line structures.
157
+
158
+ Lowered the settings.
159
+
160
+ image size: 1-10.
161
+
162
+ max_mass: 6.
163
+
164
+ # Version 22
165
+
166
+ Still with disabled `NEAREST4`, `ALL8`, `CORNER4`. So the model can be trained only on line structures.
167
+
168
+ Bigger images. image size: 1-15.
169
+
170
+ max_mass: 6.
171
+
172
+ # Version 23
173
+
174
+ Enabled all the transformations.
175
+
176
+ Bigger images. image size: 1-20.
177
+
178
+ max_mass: 8.
179
+