 Okay, we don't know, we don't have the sound from the computer. Okay, something is on the right with the sound again. Oops, try this again. My microphone is okay, so I might just keep going without the computer sound. No idea what happened to that. Clean up a lot of this stuff. I don't have any sound at all. Okay, I'll just jump into it and see how we go. Okay, so we popped in some Python code and JavaScript into GPT-4. Hey, GitHub Compound was good at removing the comments. So we removed some comments from the code just so it's less text for the website, the OpenAI website. Now this app, we were working on it yesterday. That's some issues. Leave it. Of course, there's the text. It doesn't appear properly, but many. It's the detection. It's always outputting square meant to do a fuzzy logic detection on the signal type. Vivo Vivo noise. It's the detection error, 100%. Anyway, it's not working well at the moment. Last time we were modifying the Python code. Let's do an explain this quickly and then move to JavaScript. I think after updating Python code, we had to also update JavaScript, to last time. So yes, we have analyzed, I think it's a post request that we are using. Wait a second. Are there any templates? Template up. The template function is a flash function that generates the webpage from a template. In this case, I'm using the index HTML template. Okay, where's the index HTML? It should be working. Okay. Know how it has a context or not? GPT definitely does. It really loaded the index HTML and the metadata stunning main content graphs. It's okay. This is the flask of Python code. Python script, flask application, road signal generation, feature extraction, error, running the app. And there is a JavaScript code as well. We removed any comments from it for now. Many. So we started this just yesterday. Many of the templates in the code that will have to complete or remove or update the template. And that's right. So we have the part in the copilot. We have the Python HTML. I don't quite understand how the context works for GitHub Copilot. Oh, GPT4 is working better. So we have that up to the output. And no percentage. Evadlessness. Yes. I want to try also slash workspace. What's the structure? It gives me a red buff allocation fail. I think it's too much for it to handle. Are there any templates in the code provided above? Yep. Now that's not the type of template I meant. Let's talk about the HTML template. I meant any missing script. I removed the comments from this code just so I can. So it's easier to load it into GPT4. Which says that he has the flask application. Yeah, that's not a template. I mean any... Sorry, they're not called templates. They're called placeholders. Let's correct that. Placeholders. The JavaScript seem to be complete. Plotly library input wave. That was amplitude. Yeah, those are IDs from HTML. Analyze endpoint. No, it doesn't work as expected. Any placeholders in the code. Same question about the... Because in GitHub, you have to essentially work with each file separately. GPT4, an open-air website able to hold much longer context. JavaScript code. Function a bit to be complete. But maybe consider placeholder. Well, yeah, let's check that one quickly. So it will be a generate signal. It takes the input wave as the text. And currently has the square triangle left drop zone. Right, so it's missing stuff. I'm sure I did add this yesterday. I meant to have all these other functions square. We actually don't have a triangle. We do have a left drop zone. But everything else results to... Goes to default, sine wave. So that's not cool. So we have to update this thingies. So we have those functions in Python code. Signal push. Okay. So in Python code, we actually define the actual wave form. A generate signal function to actually have a hexagon. Yes, we do. And in the output, we should have also an unknown. Let's see if GitHub Copilot can just fix this for us. Because one thing, we don't have a triangle. Switch input case. Left drop zone should be the same. We actually have a triangle. We can get rid of it as well. Do we have a triangle? I don't know. Okay. I'll remember it. Left drop zone. Sine is default. That still doesn't work. Anyway, they might not be doing much development today. The idea is for you to select the input wave. It does a fuzzy logic, which we also need to fix. Essentially have three membership. A function based on peak sharpness and symmetry. And then meant in the output here, detect the input wave form. Based on just looking at some dark points and feeding them into the fuzzy logic. Threshold set. And it also will give you a detection error rate and the amount of unknowns. So when the noise is much larger, it would not be able to detect the original signal. So we didn't have much time today. We'll continue this tomorrow. Join in. Have fun. See you later.