-
Notifications
You must be signed in to change notification settings - Fork 0
/
datapoint.go
54 lines (43 loc) · 1.43 KB
/
datapoint.go
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
package datasets
import (
"fmt"
"github.com/EganBoschCodes/lossless/neuralnetworks/save"
)
type DataPoint struct {
Input []float64
Output []float64
}
func (dp *DataPoint) ToBytes() []byte {
return append(save.ToBytes(dp.Input), save.ToBytes(dp.Output)...)
}
func DataPointFromBytes(bytes []byte, inputLength int) DataPoint {
values := save.FromBytes(bytes)
return DataPoint{Input: values[:inputLength], Output: values[inputLength:]}
}
func SaveDataset(dataset []DataPoint, dir string, name string) {
bytes := save.ConstantsToBytes(len(dataset[0].Input), len(dataset[0].Output))
for _, dp := range dataset {
bytes = append(bytes, dp.ToBytes()...)
}
if len(dir) > 0 {
save.WriteBytesToFile(fmt.Sprintf("%s/%s.dtst", dir, name), bytes)
} else {
save.WriteBytesToFile(fmt.Sprintf("%s.dtst", name), bytes)
}
}
func OpenDataset(dir string, name string) []DataPoint {
var rawBytes []byte
if len(dir) > 0 {
rawBytes = save.ReadBytesFromFile(fmt.Sprintf("%s/%s.dtst", dir, name))
} else {
rawBytes = save.ReadBytesFromFile(fmt.Sprintf("%s.dtst", name))
}
metadataRaw, datapointsRaw := rawBytes[:8], rawBytes[8:]
metadata := save.ConstantsFromBytes(metadataRaw)
inputLength, stride := metadata[0], (metadata[0]+metadata[1])*8
dataset := make([]DataPoint, len(datapointsRaw)/stride)
for i := 0; i < len(datapointsRaw); i += stride {
dataset[i/stride] = DataPointFromBytes(datapointsRaw[i:i+stride], inputLength)
}
return dataset
}