tx · EPzQDj6V1WSZyt31d5JN4yQdfxqisDnUmXHEcK9HZnRU

3N3n75UqB8G1GKmXFr4zPhKCjGcqJPRSuJY:  -0.01000000 Waves

2024.05.24 17:48 [3120229] smart account 3N3n75UqB8G1GKmXFr4zPhKCjGcqJPRSuJY > SELF 0.00000000 Waves

{
    "type": 13,
    "id": "EPzQDj6V1WSZyt31d5JN4yQdfxqisDnUmXHEcK9HZnRU",
    "fee": 1000000,
    "feeAssetId": null,
    "timestamp": 1716562138680,
    "version": 2,
    "chainId": 84,
    "sender": "3N3n75UqB8G1GKmXFr4zPhKCjGcqJPRSuJY",
    "senderPublicKey": "2AWdnJuBMzufXSjTvzVcawBQQhnhF1iXR6QNVgwn33oc",
    "proofs": [
        "2AcwcJTf1aoUBeYJ8D81KngcwnszXFYx3TAUziSW4nq86rSh7yLu1SVdexFYNDTZJU8tjsauB4aEGgjqoJBgi8UV"
    ],
    "script": "base64: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",
    "height": 3120229,
    "applicationStatus": "succeeded",
    "spentComplexity": 0
}

View: original | compacted
Prev: 9o2vqzbv5JjJKboqkHfGhnn2XCjoXwLe8EaH4ffdKaSu
Next: 5iNUFVUiuLxLyD9zsZtxLmY1BS5THCkkmFMT6kNw969S

Full: 
Old
New
Differences
1
-
{-# STDLIB_VERSION 5 #-}
1
+
{-# STDLIB_VERSION 7 #-}
2
2
 
{-# SCRIPT_TYPE ACCOUNT #-}
3
3
 
{-# CONTENT_TYPE DAPP #-}
4
-
let layer1Weights = [[4496, -6718], [36738, -36738], [3517, -3496], [-39880, 39880]]
4
+
let weights_layer_1 = [[60050, 60073], [41420, 41425]]
5
5
 
6
-
let layer1Biases = [18719, -1, 29077, 0]
6
+
let biases_layer_1 = [-25905, -63564]
7
7
 
8
-
let layer2Weights = [[-11902, 64358, -26901, 48391]]
8
+
let weights_layer_2 = [[83297, -89714]]
9
9
 
10
-
let layer2Biases = [-6873]
10
+
let biases_layer_2 = [-38118]
11
11
 
12
-
func relu (x) = if ((x > 0))
13
-
    then x
14
-
    else 0
15
-
16
-
17
-
func sigmoid_approx (x) = if ((-4000 > x))
18
-
    then 0
19
-
    else if ((-2000 > x))
20
-
        then 1000
21
-
        else if ((-1000 > x))
22
-
            then 2000
23
-
            else if ((0 > x))
24
-
                then 3000
25
-
                else if ((1000 > x))
26
-
                    then 5000
27
-
                    else if ((2000 > x))
28
-
                        then 7000
29
-
                        else if ((3000 > x))
30
-
                            then 8000
31
-
                            else if ((4000 > x))
32
-
                                then 9000
33
-
                                else 10000
34
-
35
-
36
-
func dotProduct (v1,v2) = (((v1[0] * v2[0]) + (v1[1] * v2[1])) / 10000)
37
-
38
-
39
-
func feedforward (inputs) = {
40
-
    let dp1 = (dotProduct(inputs, layer1Weights[0]) + layer1Biases[0])
41
-
    let dp2 = (dotProduct(inputs, layer1Weights[1]) + layer1Biases[1])
42
-
    let dp3 = (dotProduct(inputs, layer1Weights[2]) + layer1Biases[2])
43
-
    let dp4 = (dotProduct(inputs, layer1Weights[3]) + layer1Biases[3])
44
-
    let layer1Results = [relu(dp1), relu(dp2), relu(dp3), relu(dp4)]
45
-
    let dpLayer2 = (dotProduct(layer1Results, layer2Weights[0]) + layer2Biases[0])
46
-
    let output = sigmoid_approx(dpLayer2)
47
-
    $Tuple6(output, dp1, dp2, dp3, dp4, dpLayer2)
12
+
func linear_forward (input,weights,biases) = {
13
+
    let weighted_sum1 = ((((input[0] * weights[0][0]) + (input[1] * weights[0][1])) / 10000) + biases[0])
14
+
    let weighted_sum2 = ((((input[0] * weights[1][0]) + (input[1] * weights[1][1])) / 10000) + biases[1])
15
+
[weighted_sum1, weighted_sum2]
48
16
 
    }
49
17
 
50
18
 
19
+
func relu (input) = {
20
+
    let out1 = if ((input[0] > 0))
21
+
        then input[0]
22
+
        else 0
23
+
    let out2 = if ((input[1] > 0))
24
+
        then input[1]
25
+
        else 0
26
+
[out1, out2]
27
+
    }
28
+
29
+
30
+
func sigmoid (input) = if ((-10000 > input))
31
+
    then 0
32
+
    else if ((input > 10000))
33
+
        then 10000
34
+
        else (5000 + (input / 2))
35
+
36
+
51
37
 
@Callable(i)
52
-
func predict (input1,input2) = {
53
-
    let inputs = [input1, input2]
54
-
    let $t015731641 = feedforward(inputs)
55
-
    let prediction = $t015731641._1
56
-
    let dp1 = $t015731641._2
57
-
    let dp2 = $t015731641._3
58
-
    let dp3 = $t015731641._4
59
-
    let dp4 = $t015731641._5
60
-
    let dpLayer2 = $t015731641._6
61
-
[IntegerEntry("prediction", prediction), IntegerEntry("dotProduct1", dp1), IntegerEntry("dotProduct2", dp2), IntegerEntry("dotProduct3", dp3), IntegerEntry("dotProduct4", dp4), IntegerEntry("dotProductLayer2", dpLayer2)]
38
+
func predict (x1,x2) = {
39
+
    let inputs = [(x1 * 10000), (x2 * 10000)]
40
+
    let z1 = linear_forward(inputs, weights_layer_1, biases_layer_1)
41
+
    let a1 = relu(z1)
42
+
    let debug_z1_1 = IntegerEntry("debug_z1_1", z1[0])
43
+
    let debug_z1_2 = IntegerEntry("debug_z1_2", z1[1])
44
+
    let debug_a1_1 = IntegerEntry("debug_a1_1", a1[0])
45
+
    let debug_a1_2 = IntegerEntry("debug_a1_2", a1[1])
46
+
    let z2 = ((((a1[0] * weights_layer_2[0][0]) + (a1[1] * weights_layer_2[0][1])) / 10000) + biases_layer_2[0])
47
+
    let a2 = sigmoid(z2)
48
+
    let result = (a2 / 10000)
49
+
    let debug_z2 = IntegerEntry("debug_z2", z2)
50
+
    let debug_a2 = IntegerEntry("debug_a2", a2)
51
+
    let debug_result = IntegerEntry("debug_result", result)
52
+
    $Tuple2([debug_z1_1, debug_z1_2, debug_a1_1, debug_a1_2, debug_z2, debug_a2, debug_result], result)
62
53
 
    }
63
54
 
64
55
 

github/deemru/w8io/c3f4982 
38.32 ms ◑