 So we're focusing around this CG spectrogram spectrum. We have this app, a flat complication that looks like this. We'll actually execute. Yes. You can select your channel, change the window size. That doesn't work. Turning the training on and off. Well, main problem is that we do not see anything. Data channel data is undefined. Why? Yeah, if the filter order is too high, we'll get that as well. Okay, filter, filter order is fine. What's the default value? Zero? It's not zero. Yeah, five. Have any errors? We have the data displayed. Yeah, there's a triple click. A whole spectrum is actually not showing anything. Right. We need a developer. A software developer. A full stack. Yeah, chgpt4. Think default. And we normally start. So we have those prompts, prompts, prompts, prompts. We used them before. Start with this code. Right. So we started. We had the beginning of a code that another bot actually created. So we have the continuation of that. You can always detail that we are using a color scheme, but that's a bit messy. I have to clean it up. We can do it later. Yeah, that filter order will actually change the default processing, which has not showing any signal. A window size input is not hitting the charts. Also, yeah, whole spectrum is not working properly. How did you know? I'm not sure about the frequency scale. Yeah, frequency scale. If we don't have the play button, we don't have the scroller as well. File, scroller is missing. A make sure second chart has frequency as x-axis and power as y-axis. A raw signature is not showing any signal. Also make sure the chart has time as x-axis and the true as y-axis. Okay, so we have those. A prompt play button is missing as well. Right. So let's start with index HTML. Let's generate for a bit just to make sure it's actually read the thing. The JavaScript. Okay, stop generating the JavaScript and make sure last application Python code. Okay, we can stop generating that. This is the third pile, the Python code still generating. Okay, we can stop that. Now we have our prompts with all the issues that we are having. This should be plenty by now. This is what it currently looks like. There's a lot of problems with that. Also make sure the linear logarithmic scale should be applied. Yeah, the file scroll is updating the charts. That's a good idea. Make sure log a linear scale is changing meant to be the x-axis of the second chart. That's right. So we have a bunch of prompts ready to pop in. And yes, we're celebrating 100 years will be in there. So the first human EEG was recorded by Hans Berger back in 1924. So we're celebrating that next year. By the way, the more bands we're currently doing only alpha, beta, theta. But there's also gamma, just pretty much anything above 32 hertz and also mu between 8 and 12. It's the same as alpha. I felt it differently or something. I don't know. But yeah, alpha, beta, delta, theta should be a good start. Several issues. Yeah, didn't generate any code. Yeah, I would like you to hold my hand and give me step by step guidance. So essentially it's saying too much, too many prompts at once in the correct time. And amplitude data. So it's a good start. Can process EEG function in Python? Have a process EEG function somewhere over here. A filtered data right over here using the field field function, converting to least the forward, backward, basic filter. This is the same. Yes. So why did you mention it? When returning the JSON, JSON file, we have filtered data, whole spectrum. So filter data has a channel number BA, right? We're already doing a Butterworth filter. It's a band pass, right? Those are the coefficients and then the coefficients are being used in here. So that should be fine. Filter order is taken from here. No, from there. It's suggesting that we are not passing the right values. So instead of that into JSON, we want to pass this. Doesn't make sense though. Can you explain this? Can we use co-pilot for that? It's a generic. Can co-pilot do this? Filter data in other bands, front end. This is in JavaScript. I think this makes more sense. Okay, so now it's displaying something. This is obviously wrong. Should be starting off a filter of delta is unstable. That's normally when we need to decrease the order of the filter. So now it doesn't say unstable anymore, but it's also not showing anything useful. Okay, so we haven't also finished modifying the code. So we have this time. That's correct. So we need to parse channel number to integer. Pretty sure it's already an integer. So we're already doing that. We need to replace channel number. Get element by integer. Number wouldn't be an integer already. I don't know what it's talking about. Verify that we're receiving correct data. Yeah, console log is not a bad idea. Let's just see if turning delta is unstable. What? So there are no nouns before we had a problem with nouns. Getting a channel that is undefined. We show we have this already showing some data. What's this problem? Yes, but a common HTML container called raw signal. It does exist. Now let's try something unusual. Let's try this. Let's try print screen straight into GPT-4. This is the front end to fix. Yeah, that's what I thought. That's probably not the case. Now there is nothing when zooming in. Where's the top? It's still giving me that undefined. It's giving me some data. Okay, let's try co-pilot. Let's go back to the window. This is raw signal. JS, raw signal, plot, raw signal data. It's appearing twice. It's giving the same thing. I don't know what's up. Am I plotting that? I don't know. Those essentially are empty, are they? The object has the proper field of data, and if the filtered operator has an array element at the index. So this should be true unless there's a better explanation of this. Yeah, there's a confusion with the names. When calling without any conditions, this is plotly. This can be replaced. First, in the HTML, we need these values smaller. Let's do a second. Let's plot raw signal. Why am I undefined? Check the data source. Raw signal is just an example. What is the data structure? So this filtered data has a different bands. You should have the whole thing as well. It's a whole spectrum data. Can you fix this for us? I don't know how to do it. I'm just going to write it. It's already there. The filtered data would not have a channel number. Okay, let's try this again. Can channel number be passed on in another way? I don't understand. Why is it not able to fix the problem? Let it run. First of all, we're having an issue in JavaScript in line 29. Yeah, we're having a problem with this server client data exchange. Yeah, why is the channel number in the filtered data should be there? Yeah, I know that's what I have, but that's the problem with JSON. There's always strings. Invite channel current channel data. We're already creating it to list. Doesn't make any sense at all. Okay, currently what we have is not exactly working. Make sure it's showing signal and make sure window sizes would is dating the charts. All spectrum is working properly and file scroller is in boot. Make sure file scroll is updating the charts. Make sure the log limit scale is changing the X-axis of the second chart. Do not generate any code yet because it won't be able to do it in one go. We'll ask it to regenerate the files one by one and make sure no missing bits. So place all the original code that remains the same. Okay, let's do index HTML first. What's up with that? I can change this. There would be additional modification in function to add a file. It's not file input. It's file scroller. No, there's only one file. File scroller. It's a horizontal file scroller to move through the file again. From here, something went wrong. Here's the plan. Why does the variable contains the correct channel? That would be nice. Window size, second chart. It should be similar to this but with quite a few changes. We should have a file scroller like that. Play button scroll during the charts. Horizontal scroller moves through the file. Scroller slider sounds good. Yeah, we'll need to update the CSS. Come on, why are you ignoring my instructions? It's still generating. Make sure to include all necessary code. I'll hit the limit, won't I? What's it now? 50 prompts. Yeah, a cap of 50 messages every three hours. It's meant to be moving windows. Yeah, now it seems to generate a whole code without any missing parts. It's still generating. I don't know why it stopped. Give it some more time. I'm just stuck here for the reason. What's going on? Let's ask Bud. My GPT-4 is getting stuck a lot recently. Network error. That's why. Should I regenerate, restart the window? The CSS should have been using the colors. We'll change it in sec. Now this JavaScript. The JavaScript. Last application, the Python code. It's loading something. Takes a while. Not getting any errors, but not getting any data. Let's stop this for a sec. A control shift C. The one we got an error with. So this is the Python code. Now we had an update to the JavaScript. It wasn't complete. And now when it's not complete, it said no placeholders. I guess we're doing it several times. It kind of helps. Let's start again. Shall we? What we currently have isn't working. So default GPT-4. We have this first create. Last update. In the time and frequency domains. The app should have a fast growing play button. Frequency scale. I'm sure about the frequency scale selector. Should also have a filter order. That's crazy. It's crazy long romped. Standing in outline. And forget about the requirements file. Instead, do not generate any code. Okay. And we want the bot to keep generating. Creating API rot-fetch. Update the database to the input. There's a post request that makes sense. So templates is fine. A lot of to-dos. Does it look like? We got the HTML in the app. From E-G-J-S. It's such a script. Now the style CSS. We want it in these colors. These variables. Let's see what we got. What's that? Okay. It's not much. Go through the whole romped. But it ignored a whole heap of stuff. A HTML would look something like this. It's pretty basic. We'll expect it to window sizes. Okay. So now it's using JSON. Now I keep changing its mind. As if it has a mind. Assume it does. Because that's what it was suggesting me doing. The charts are not showing anything. Just keep it nice and simple. Had a better version of it already. But I got too complicated. 51 in J-S generic stuff. So we do have this error. We have channel selector. Use copilot. Populating that. 16 channels. Can we do it differently? Right. Start button. No, there's no stop button. Okay, let's go back to GPT. Instead of that. Index HTML. Update. Regenerate the whole code. See what it does. What? Do I not ask for index HTML? Right. That's generally HTML. Okay. So this would be the same. Obviously. Channel number. Window size. Ideally those would be scrollers, but that's okay. Why it's refusing just to do the whole lot? Why not just complete the code? Still generating. I do want the signal, obviously. We already have an ID there. Yeah, plot is fine. Generate. Oh, P-G-J-S. I don't think you can do it. Something can change. It's not generating all files. Fetch data. Yet I'm not sure frequency power. That's correct. That's exactly what I want. That looks much better. Still not work. So we don't have the JSON file. We want the whole pie. Okay, I can only assume it's working better when it's nighttime in the United States or something. Because it's actually generating complete the code. Is it shorter than what I had before? JSON file. Fine consta. Stuffed up the file. Looks legit. Changed. Meant to be capital. Can read of capital letter. And this looks pretty basic. Get data. Sold the same. Now we do want FS. It's actually FS in non-capital. Letters. Get data. Name signal is not defined. Yep, we reached the current cap. Yeah, I thought I would reach it much sooner. I only reached it like in two hours. So I did more than 30 prompts. What's it, 50 prompts per three hour rate? Now it's giving me a timeout of almost one hour.