# Jarvis

So i was looking at the csv and thought `"kek tf does this mean"` but then i checked the range values for the csv and saw they were rly similar. i looked at the hint about tony stark and thought maybe i need to build an ai or some shit. woa kept telling me i was mental and had big confusion but i continued anyway. i built one (code below) and ran it (followed tutorial with adjustments for massive training data). the output it gave was in binary (like in training data). i converted to text and gave `flag{mlWis_cool}` but it looked like some bits were flipped so i changed it to `flag{ml_is_cool}` which is flag.

<https://medium.com/technology-invention-and-more/how-to-build-a-simple-neural-network-in-9-lines-of-python-code-cc8f23647ca1>

```python
hlp = open("help.csv","r").read().split("\n")
hlp = [[int(y) for y in x.split(",")] for x in hlp]

flg = open("flag.csv","r").read().split("\n")
flg = [[int(y) for y in x.split(",")] for x in flg]


from numpy import exp, array, random, dot, set_printoptions, inf
set_printoptions(threshold=inf)

class NeuralNetwork():
    def __init__(self):
        self.synaptic_weights = 2 * random.random((10, 1)) - 1
    def __sigmoid(self, x):
        return 1 / (1 + exp(-x))
    def __sigmoid_derivative(self, x):
        return x * (1 - x)
    def train(self, training_set_inputs, training_set_outputs, number_of_training_iterations):
        for iteration in range(number_of_training_iterations):
            output = self.think(training_set_inputs)
            error = training_set_outputs - output
            adjustment = dot(training_set_inputs.T, error * self.__sigmoid_derivative(output))
            self.synaptic_weights = self.synaptic_weights + adjustment # fukin numpy being shit kek
    def think(self, inputs):
        return self.__sigmoid(dot(inputs, self.synaptic_weights))



neural_network = NeuralNetwork()

print("Random starting synaptic weights: ")
print(neural_network.synaptic_weights)
training_set_inputs = [[y/100 for y in x[1:]] for x in hlp]
training_set_outputs = [x[0] for x in hlp]

print(training_set_inputs[0],training_set_outputs[0])

for i,j in enumerate(training_set_inputs):
    neural_network.train(array([j]), array(training_set_outputs[i]).T, 10000) # gotta train individually or numpy gets triggered

print("New synaptic weights after training: ")
print(neural_network.synaptic_weights)

b = ""
for i,j in enumerate(flg):
    print("Considering new situation :",j)
    a = neural_network.think(array(flg[i]))[0]
    b += str(int(a))
    print(int(a))
print(b)
```


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://the-winrars.gitbook.io/writeups/2020-writeups/tjctf/misc/jarvis.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.