tx · GtG7mCapTmYceaRkKY1vhdnd2yLu3vXtwL6gnTomt2N9

3N3n75UqB8G1GKmXFr4zPhKCjGcqJPRSuJY:  -0.01000000 Waves

2024.04.28 14:25 [3082655] smart account 3N3n75UqB8G1GKmXFr4zPhKCjGcqJPRSuJY > SELF 0.00000000 Waves

{
    "type": 13,
    "id": "GtG7mCapTmYceaRkKY1vhdnd2yLu3vXtwL6gnTomt2N9",
    "fee": 1000000,
    "feeAssetId": null,
    "timestamp": 1714303469498,
    "version": 2,
    "chainId": 84,
    "sender": "3N3n75UqB8G1GKmXFr4zPhKCjGcqJPRSuJY",
    "senderPublicKey": "2AWdnJuBMzufXSjTvzVcawBQQhnhF1iXR6QNVgwn33oc",
    "proofs": [
        "21SPdzhzPPKjDKz7mSUHHX6qmiQY9Ucrh6vTU8dNKvwpjvFJ1SmHyx6K5H2Z7THDSECMmeLa6VuPLj2sKdXJEon6"
    ],
    "script": "base64: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",
    "height": 3082655,
    "applicationStatus": "succeeded",
    "spentComplexity": 0
}

View: original | compacted
Prev: 8Tiq9PrpLGFMXBGGABe7MVibryxyyrcnbvTnKbFxbEWk
Next: CdzzpSdtWkMdq88oLky7G7s1vveHjZQsJ1rC8WaAeC1X

Diff: 
Old
New
Differences
1
1
 
{-# STDLIB_VERSION 5 #-}
2
2
 
{-# SCRIPT_TYPE ACCOUNT #-}
3
3
 
{-# CONTENT_TYPE DAPP #-}
4
-
let layer1Weights = [[600496, 600733], [414197, 414253]]
4
+
let layer1Weights = [[600497, 600732], [414197, 414253]]
5
5
 
6
6
 
let layer1Biases = [-259051, -635637]
7
7
 
9
9
 
10
10
 
let layer2Biases = [-381179]
11
11
 
12
-
func clampZ (z,limit) = if ((z > limit))
13
-
    then limit
14
-
    else if ((-(limit) > z))
15
-
        then -(limit)
16
-
        else z
17
-
18
-
19
-
func exp_approx (x) = {
20
-
    let abs_x = if ((0 > x))
21
-
        then -(x)
22
-
        else x
23
-
    let adjusted_x = if ((abs_x > 100000))
24
-
        then 100000
25
-
        else abs_x
26
-
    let scaled_x = (adjusted_x / 1000)
27
-
    (100000 - (10 * scaled_x))
12
+
func expApprox (x) = {
13
+
    let scaledX = fraction(x, 1, 1000000)
14
+
    let scaledX2 = fraction(scaledX, scaledX, 1000000)
15
+
    let term1 = (1000000 - scaledX)
16
+
    let term2 = fraction(scaledX2, 500000, 1)
17
+
    (term1 + term2)
28
18
 
    }
29
19
 
30
20
 
31
21
 
func sigmoid (z,debugPrefix) = {
32
-
    let clampedZ = clampZ(z, 100000)
33
-
    let expValue = exp_approx(-(clampedZ))
34
-
    let sigValue = fraction(1000000, (1000000 + expValue), 1, DOWN)
35
-
    $Tuple2([IntegerEntry((debugPrefix + "inputZ"), z), IntegerEntry((debugPrefix + "clampedZ"), clampedZ), IntegerEntry((debugPrefix + "expValue"), expValue), IntegerEntry((debugPrefix + "sigValue"), sigValue)], sigValue)
22
+
    let expNegZ = expApprox(-(z))
23
+
    let onePlusExpNegZ = (1000000 + expNegZ)
24
+
    let sigValue = fraction(1000000000000, onePlusExpNegZ, 1)
25
+
    $Tuple2([IntegerEntry((debugPrefix + "inputZ"), z), IntegerEntry((debugPrefix + "expNegZ"), expNegZ), IntegerEntry((debugPrefix + "onePlusExpNegZ"), onePlusExpNegZ), IntegerEntry((debugPrefix + "sigValue"), sigValue)], sigValue)
36
26
 
    }
37
27
 
38
28
 
39
29
 
func forwardPassLayer1 (input,weights,biases,debugPrefix) = {
40
30
 
    let sum0 = (((input[0] * weights[0][0]) + (input[1] * weights[0][1])) + (biases[0] * 100000))
41
31
 
    let sum1 = (((input[0] * weights[1][0]) + (input[1] * weights[1][1])) + (biases[1] * 100000))
42
-
    let $t018561909 = sigmoid(sum0, "Layer1N0")
43
-
    let debugEntries0 = $t018561909._1
44
-
    let sig0 = $t018561909._2
45
-
    let $t019141967 = sigmoid(sum1, "Layer1N1")
46
-
    let debugEntries1 = $t019141967._1
47
-
    let sig1 = $t019141967._2
32
+
    let $t013271380 = sigmoid(sum0, "Layer1N0")
33
+
    let debugEntries0 = $t013271380._1
34
+
    let sig0 = $t013271380._2
35
+
    let $t013851438 = sigmoid(sum1, "Layer1N1")
36
+
    let debugEntries1 = $t013851438._1
37
+
    let sig1 = $t013851438._2
48
38
 
    let debugInfo = (debugEntries0 ++ debugEntries1)
49
39
 
    let output = [sig0, sig1]
50
40
 
    $Tuple2(debugInfo, output)
53
43
 
54
44
 
func forwardPassLayer2 (input,weights,biases,debugPrefix) = {
55
45
 
    let sum0 = (((input[0] * weights[0][0]) + (input[1] * weights[0][1])) + (biases[0] * 100000))
56
-
    let $t022772330 = sigmoid(sum0, "Layer2N0")
57
-
    let debugEntries0 = $t022772330._1
58
-
    let sig0 = $t022772330._2
46
+
    let $t017481801 = sigmoid(sum0, "Layer2N0")
47
+
    let debugEntries0 = $t017481801._1
48
+
    let sig0 = $t017481801._2
59
49
 
    let debugInfo = debugEntries0
60
50
 
    let output = sig0
61
51
 
    $Tuple2(debugInfo, output)
71
61
 
        then 1000000
72
62
 
        else 0
73
63
 
    let inputs = [scaledInput1, scaledInput2]
74
-
    let $t026422740 = forwardPassLayer1(inputs, layer1Weights, layer1Biases, "Layer1")
75
-
    let debugLayer1 = $t026422740._1
76
-
    let layer1Output = $t026422740._2
77
-
    let $t027452849 = forwardPassLayer2(layer1Output, layer2Weights, layer2Biases, "Layer2")
78
-
    let debugLayer2 = $t027452849._1
79
-
    let layer2Output = $t027452849._2
64
+
    let $t021132211 = forwardPassLayer1(inputs, layer1Weights, layer1Biases, "Layer1")
65
+
    let debugLayer1 = $t021132211._1
66
+
    let layer1Output = $t021132211._2
67
+
    let $t022162320 = forwardPassLayer2(layer1Output, layer2Weights, layer2Biases, "Layer2")
68
+
    let debugLayer2 = $t022162320._1
69
+
    let layer2Output = $t022162320._2
80
70
 
    (([IntegerEntry("result", layer2Output)] ++ debugLayer1) ++ debugLayer2)
81
71
 
    }
82
72
 

Full: 
Old
New
Differences
1
1
 
{-# STDLIB_VERSION 5 #-}
2
2
 
{-# SCRIPT_TYPE ACCOUNT #-}
3
3
 
{-# CONTENT_TYPE DAPP #-}
4
-
let layer1Weights = [[600496, 600733], [414197, 414253]]
4
+
let layer1Weights = [[600497, 600732], [414197, 414253]]
5
5
 
6
6
 
let layer1Biases = [-259051, -635637]
7
7
 
8
8
 
let layer2Weights = [[832966, -897142]]
9
9
 
10
10
 
let layer2Biases = [-381179]
11
11
 
12
-
func clampZ (z,limit) = if ((z > limit))
13
-
    then limit
14
-
    else if ((-(limit) > z))
15
-
        then -(limit)
16
-
        else z
17
-
18
-
19
-
func exp_approx (x) = {
20
-
    let abs_x = if ((0 > x))
21
-
        then -(x)
22
-
        else x
23
-
    let adjusted_x = if ((abs_x > 100000))
24
-
        then 100000
25
-
        else abs_x
26
-
    let scaled_x = (adjusted_x / 1000)
27
-
    (100000 - (10 * scaled_x))
12
+
func expApprox (x) = {
13
+
    let scaledX = fraction(x, 1, 1000000)
14
+
    let scaledX2 = fraction(scaledX, scaledX, 1000000)
15
+
    let term1 = (1000000 - scaledX)
16
+
    let term2 = fraction(scaledX2, 500000, 1)
17
+
    (term1 + term2)
28
18
 
    }
29
19
 
30
20
 
31
21
 
func sigmoid (z,debugPrefix) = {
32
-
    let clampedZ = clampZ(z, 100000)
33
-
    let expValue = exp_approx(-(clampedZ))
34
-
    let sigValue = fraction(1000000, (1000000 + expValue), 1, DOWN)
35
-
    $Tuple2([IntegerEntry((debugPrefix + "inputZ"), z), IntegerEntry((debugPrefix + "clampedZ"), clampedZ), IntegerEntry((debugPrefix + "expValue"), expValue), IntegerEntry((debugPrefix + "sigValue"), sigValue)], sigValue)
22
+
    let expNegZ = expApprox(-(z))
23
+
    let onePlusExpNegZ = (1000000 + expNegZ)
24
+
    let sigValue = fraction(1000000000000, onePlusExpNegZ, 1)
25
+
    $Tuple2([IntegerEntry((debugPrefix + "inputZ"), z), IntegerEntry((debugPrefix + "expNegZ"), expNegZ), IntegerEntry((debugPrefix + "onePlusExpNegZ"), onePlusExpNegZ), IntegerEntry((debugPrefix + "sigValue"), sigValue)], sigValue)
36
26
 
    }
37
27
 
38
28
 
39
29
 
func forwardPassLayer1 (input,weights,biases,debugPrefix) = {
40
30
 
    let sum0 = (((input[0] * weights[0][0]) + (input[1] * weights[0][1])) + (biases[0] * 100000))
41
31
 
    let sum1 = (((input[0] * weights[1][0]) + (input[1] * weights[1][1])) + (biases[1] * 100000))
42
-
    let $t018561909 = sigmoid(sum0, "Layer1N0")
43
-
    let debugEntries0 = $t018561909._1
44
-
    let sig0 = $t018561909._2
45
-
    let $t019141967 = sigmoid(sum1, "Layer1N1")
46
-
    let debugEntries1 = $t019141967._1
47
-
    let sig1 = $t019141967._2
32
+
    let $t013271380 = sigmoid(sum0, "Layer1N0")
33
+
    let debugEntries0 = $t013271380._1
34
+
    let sig0 = $t013271380._2
35
+
    let $t013851438 = sigmoid(sum1, "Layer1N1")
36
+
    let debugEntries1 = $t013851438._1
37
+
    let sig1 = $t013851438._2
48
38
 
    let debugInfo = (debugEntries0 ++ debugEntries1)
49
39
 
    let output = [sig0, sig1]
50
40
 
    $Tuple2(debugInfo, output)
51
41
 
    }
52
42
 
53
43
 
54
44
 
func forwardPassLayer2 (input,weights,biases,debugPrefix) = {
55
45
 
    let sum0 = (((input[0] * weights[0][0]) + (input[1] * weights[0][1])) + (biases[0] * 100000))
56
-
    let $t022772330 = sigmoid(sum0, "Layer2N0")
57
-
    let debugEntries0 = $t022772330._1
58
-
    let sig0 = $t022772330._2
46
+
    let $t017481801 = sigmoid(sum0, "Layer2N0")
47
+
    let debugEntries0 = $t017481801._1
48
+
    let sig0 = $t017481801._2
59
49
 
    let debugInfo = debugEntries0
60
50
 
    let output = sig0
61
51
 
    $Tuple2(debugInfo, output)
62
52
 
    }
63
53
 
64
54
 
65
55
 
@Callable(i)
66
56
 
func predict (input1,input2) = {
67
57
 
    let scaledInput1 = if ((input1 == 1))
68
58
 
        then 1000000
69
59
 
        else 0
70
60
 
    let scaledInput2 = if ((input2 == 1))
71
61
 
        then 1000000
72
62
 
        else 0
73
63
 
    let inputs = [scaledInput1, scaledInput2]
74
-
    let $t026422740 = forwardPassLayer1(inputs, layer1Weights, layer1Biases, "Layer1")
75
-
    let debugLayer1 = $t026422740._1
76
-
    let layer1Output = $t026422740._2
77
-
    let $t027452849 = forwardPassLayer2(layer1Output, layer2Weights, layer2Biases, "Layer2")
78
-
    let debugLayer2 = $t027452849._1
79
-
    let layer2Output = $t027452849._2
64
+
    let $t021132211 = forwardPassLayer1(inputs, layer1Weights, layer1Biases, "Layer1")
65
+
    let debugLayer1 = $t021132211._1
66
+
    let layer1Output = $t021132211._2
67
+
    let $t022162320 = forwardPassLayer2(layer1Output, layer2Weights, layer2Biases, "Layer2")
68
+
    let debugLayer2 = $t022162320._1
69
+
    let layer2Output = $t022162320._2
80
70
 
    (([IntegerEntry("result", layer2Output)] ++ debugLayer1) ++ debugLayer2)
81
71
 
    }
82
72
 
83
73
 

github/deemru/w8io/026f985 
28.67 ms ◑