1 00:00:00,000 --> 00:00:00,000 Hello guys. 2 00:00:00,000 --> 00:00:06,000 So we are going to continue the discussion with respect to creating this amazing end to end search engine, 3 00:00:06,000 --> 00:00:09,000 an AI application, with the help of tools and agents. 4 00:00:09,000 --> 00:00:14,000 Already in our previous video, we have seen that how we can use Lang chain in order to create tools 5 00:00:14,000 --> 00:00:17,000 and how we can execute it with the help of agents. 6 00:00:17,000 --> 00:00:22,000 And uh, in our previous video, we have already seen about creating your own custom tools. 7 00:00:22,000 --> 00:00:27,000 Along with that, we have also seen how to use the inbuilt tools that are supported by Lang Chain. 8 00:00:27,000 --> 00:00:30,000 Now let's go ahead and quickly start our project. 9 00:00:30,000 --> 00:00:36,000 So here in my search engine folder you will be able to see that I have created my app Dot Pi project. 10 00:00:36,000 --> 00:00:40,000 Uh, so folder by folder I am trying to execute this specific project. 11 00:00:40,000 --> 00:00:42,000 So let's go ahead without wasting any time. 12 00:00:42,000 --> 00:00:48,000 So first of all we will go ahead and import Streamlit as st along with this. 13 00:00:48,000 --> 00:00:52,000 Uh, what we are basically going to do is that we are also going to import all the other things that 14 00:00:52,000 --> 00:00:53,000 we require. 15 00:00:53,000 --> 00:00:56,000 Okay, so I'll just remove this. 16 00:00:56,000 --> 00:01:00,000 I don't know why this suggestion usually comes, but can't help. 17 00:01:00,000 --> 00:01:08,000 Then uh, quickly I will go ahead and use something called as lang chain underscore grok. 18 00:01:08,000 --> 00:01:12,000 We are going to use grok over here to access the open source models. 19 00:01:12,000 --> 00:01:14,000 So I'm just going to go ahead and write Chad Grok. 20 00:01:14,000 --> 00:01:22,000 And then from this I'm also be using lang chain underscore uh, community dot utils. 21 00:01:22,000 --> 00:01:23,000 Okay. 22 00:01:23,000 --> 00:01:28,000 I've already shown you how to probably go ahead and create your wrappers. 23 00:01:28,000 --> 00:01:28,000 Right. 24 00:01:28,000 --> 00:01:31,000 So here also we are going to use the same wrappers. 25 00:01:31,000 --> 00:01:33,000 So I'll go ahead and write utilities import. 26 00:01:33,000 --> 00:01:35,000 One is the RCF wrapper. 27 00:01:35,000 --> 00:01:38,000 And then I'm going to use the Wikipedia wrapper. 28 00:01:38,000 --> 00:01:42,000 So let's go ahead and use this two wrappers for right now. 29 00:01:42,000 --> 00:01:44,000 And again you can go ahead and explore other tools. 30 00:01:44,000 --> 00:01:47,000 Also in this I will also add one search engine tool. 31 00:01:47,000 --> 00:01:48,000 Okay. 32 00:01:48,000 --> 00:01:49,000 Let let's see that also. 33 00:01:49,000 --> 00:01:50,000 Okay. 34 00:01:50,000 --> 00:01:56,000 Along with this you will be seeing that I will go ahead and import from lang chain underscore community 35 00:01:56,000 --> 00:01:57,000 dot tools. 36 00:01:57,000 --> 00:02:02,000 Inside the tools we will go ahead and import our chef query. 37 00:02:02,000 --> 00:02:06,000 And then I'll also have this uh Wikipedia query run. 38 00:02:06,000 --> 00:02:06,000 Okay. 39 00:02:06,000 --> 00:02:10,000 So our query run, and then I'll also be having this Wikipedia query run. 40 00:02:10,000 --> 00:02:16,000 Along with this I will also add one like capabilities to search from the internet. 41 00:02:16,000 --> 00:02:16,000 Right. 42 00:02:16,000 --> 00:02:22,000 So for that I have this duck duck duck duck duck go search run okay. 43 00:02:22,000 --> 00:02:28,000 So I'll be using this also which will actually help you to search anything in the internet. 44 00:02:28,000 --> 00:02:28,000 Right. 45 00:02:28,000 --> 00:02:34,000 So all these, uh, are the important imports that we are specifically going to do now, I will also 46 00:02:34,000 --> 00:02:37,000 be importing something called as from Lang Chain. 47 00:02:37,000 --> 00:02:40,000 I will be importing something called as agents. 48 00:02:40,000 --> 00:02:45,000 So let's first of all go ahead and initialize the agent over here. 49 00:02:45,000 --> 00:02:50,000 Along with this there is an option of putting something called as agent type which I will be discussing. 50 00:02:50,000 --> 00:02:51,000 What is the importance of this? 51 00:02:51,000 --> 00:02:54,000 I will go ahead and discuss about this also along with this. 52 00:02:54,000 --> 00:02:57,000 So let's go ahead and import our next library. 53 00:02:57,000 --> 00:03:01,000 So here we are also going to use something called as lang sched dot callbacks. 54 00:03:01,000 --> 00:03:06,000 And I'm going to import something called as streamlet callback handler. 55 00:03:06,000 --> 00:03:11,000 And this is important because this will allow you to communicate with all this kind of tools within 56 00:03:11,000 --> 00:03:12,000 themselves. 57 00:03:12,000 --> 00:03:12,000 Right. 58 00:03:12,000 --> 00:03:19,000 So all this libraries will be using initially right now uh streamlet has also been imported. 59 00:03:19,000 --> 00:03:23,000 I will go ahead and import OS if I want over here. 60 00:03:23,000 --> 00:03:31,000 Along with this, I will go ahead and write import OS, then from dot env import load underscore dot 61 00:03:31,000 --> 00:03:32,000 env. 62 00:03:32,000 --> 00:03:34,000 I'm also going to use this okay. 63 00:03:34,000 --> 00:03:37,000 Uh, if you remember in our chatbot also right. 64 00:03:37,000 --> 00:03:42,000 We specifically did not even have to call any APIs. 65 00:03:42,000 --> 00:03:42,000 Right. 66 00:03:42,000 --> 00:03:47,000 I was directly copying and pasting in the, uh, web application itself. 67 00:03:47,000 --> 00:03:49,000 So I will try to also develop in that specific way. 68 00:03:49,000 --> 00:03:56,000 So if I go ahead and open this OpenAI chat bot here, you'll be able to see I had created this all slide 69 00:03:56,000 --> 00:03:56,000 bars. 70 00:03:56,000 --> 00:03:56,000 Right. 71 00:03:56,000 --> 00:03:59,000 So here I will also use one of the slide bar okay. 72 00:03:59,000 --> 00:04:03,000 And I will keep this handy when I will be using it okay. 73 00:04:03,000 --> 00:04:10,000 So I'll just paste it over here to keep this sidebar settings and I'll say hey, enter your grok API 74 00:04:10,000 --> 00:04:11,000 key okay. 75 00:04:11,000 --> 00:04:14,000 So I will go ahead and enter the grok API key. 76 00:04:14,000 --> 00:04:20,000 So I'll keep this also handy so that I don't have to even call my load underscore dot EMV right for 77 00:04:20,000 --> 00:04:21,000 the same thing. 78 00:04:21,000 --> 00:04:27,000 Now, the next thing, uh, after this, what I am actually going to do is that after importing all 79 00:04:27,000 --> 00:04:31,000 these things, let's also, you know, take this entire things. 80 00:04:31,000 --> 00:04:38,000 And first of all, I will go ahead and, you know, create my RCF wrapper, API wrapper and uh, the 81 00:04:38,000 --> 00:04:39,000 Wikipedia API wrapper. 82 00:04:39,000 --> 00:04:41,000 And already I have actually shown you how to do that. 83 00:04:41,000 --> 00:04:49,000 This is for my RCF tools, which I had actually shown you in my previous video, Agip and Wikipedia 84 00:04:49,000 --> 00:04:49,000 tools. 85 00:04:49,000 --> 00:04:50,000 Right. 86 00:04:50,000 --> 00:04:56,000 So here we have initialized it, uh, along with this, as I said that I'm also going to use, uh, 87 00:04:56,000 --> 00:04:57,000 a duck. 88 00:04:57,000 --> 00:04:58,000 Duck go search. 89 00:04:58,000 --> 00:05:00,000 So that is what I'm actually going to do. 90 00:05:00,000 --> 00:05:05,000 In order to use that particular tool, I will go ahead and create a variable which is called as search 91 00:05:05,000 --> 00:05:08,000 is equal to duck, duck, go, search, run. 92 00:05:08,000 --> 00:05:13,000 And here the kind of task that I'm telling it to do, or the name that I'm actually going to give, 93 00:05:13,000 --> 00:05:15,000 this is something called as search. 94 00:05:15,000 --> 00:05:18,000 So here the name will be Wikipedia. 95 00:05:18,000 --> 00:05:23,000 Here the name will be RCF and And this will be my search functionality, which I really need to search 96 00:05:23,000 --> 00:05:25,000 from the internet itself. 97 00:05:25,000 --> 00:05:25,000 Right. 98 00:05:25,000 --> 00:05:31,000 And uh, after doing all these things now let's go ahead and create my Streamlit app. 99 00:05:31,000 --> 00:05:33,000 So first of all, I will go quickly. 100 00:05:33,000 --> 00:05:35,000 Go ahead and write s t dot title. 101 00:05:35,000 --> 00:05:39,000 I'll just go ahead and write link chain with chat with search. 102 00:05:39,000 --> 00:05:40,000 Okay. 103 00:05:40,000 --> 00:05:47,000 Um, let me just give a simple reminder or a comment over here saying that, hey, in this example, 104 00:05:47,000 --> 00:05:55,000 we are using something called as Streamlet callback handler to display the thoughts and action of the 105 00:05:55,000 --> 00:05:57,000 agent in an interactive Streamlet app. 106 00:05:57,000 --> 00:06:02,000 Okay, so this will actually help you to, uh, this Streamlet callback handler will actually help you 107 00:06:02,000 --> 00:06:06,000 to understand, like what thoughts and action of the agents will be when they are executing all this 108 00:06:06,000 --> 00:06:09,000 kind of tools, or when they are interacting with the tools. 109 00:06:09,000 --> 00:06:13,000 Okay then, I have imported this specifically for my grok API. 110 00:06:13,000 --> 00:06:14,000 Okay. 111 00:06:14,000 --> 00:06:19,000 Now what I'm actually going to do, since I also need to make sure that now my entire conversation over 112 00:06:19,000 --> 00:06:22,000 here should happen along with the chat history. 113 00:06:22,000 --> 00:06:25,000 So I will go ahead and create one simple. 114 00:06:25,000 --> 00:06:30,000 I'll just say, hey, if message is I'll, I'll basically create my session state. 115 00:06:30,000 --> 00:06:34,000 So I'll go ahead and write session underscore state over here. 116 00:06:34,000 --> 00:06:40,000 And along with this I'll write hey s t dot session state of messages okay. 117 00:06:40,000 --> 00:06:43,000 So I'm just going to go ahead and create this messages equal to. 118 00:06:43,000 --> 00:06:50,000 And here I'll just go ahead and give some of the important uh, message, uh roles that I'm actually 119 00:06:50,000 --> 00:06:51,000 going to provide. 120 00:06:51,000 --> 00:06:55,000 So first one information that I really need to put is something about role. 121 00:06:55,000 --> 00:06:57,000 So role will be nothing but assistant. 122 00:06:58,000 --> 00:07:01,000 And then I will say, hey, this is my content. 123 00:07:01,000 --> 00:07:03,000 So let me just go ahead and write the content. 124 00:07:03,000 --> 00:07:15,000 It will be again colon I'll say, hi, I am a chat bot who can search the web. 125 00:07:15,000 --> 00:07:16,000 Okay. 126 00:07:17,000 --> 00:07:19,000 Uh, or I'll also put one message. 127 00:07:19,000 --> 00:07:20,000 How can I help you? 128 00:07:20,000 --> 00:07:26,000 Okay, so this will be the default assistant, uh, information that you have. 129 00:07:27,000 --> 00:07:28,000 So this. 130 00:07:28,000 --> 00:07:30,000 I'm saving it in the session state. 131 00:07:30,000 --> 00:07:31,000 Okay. 132 00:07:31,000 --> 00:07:35,000 Then I will go ahead and say for every messages, we will go ahead and traverse it. 133 00:07:35,000 --> 00:07:45,000 So I'll write for message in s t dot session state session s t dot session. 134 00:07:46,000 --> 00:07:49,000 Underscore state dot messages. 135 00:07:49,000 --> 00:07:50,000 Okay. 136 00:07:50,000 --> 00:07:55,000 And here I'm just going to use state dot chat underscore message. 137 00:07:55,000 --> 00:08:00,000 I'm basically going to give my message role. 138 00:08:02,000 --> 00:08:03,000 Okay. 139 00:08:04,000 --> 00:08:12,000 Now inside this message role I'm just going to go ahead and write dot write right dot right over here 140 00:08:12,000 --> 00:08:14,000 as my message content. 141 00:08:14,000 --> 00:08:20,000 So with respect to any conversation that I get because I'm going to put all this specific role information 142 00:08:20,000 --> 00:08:24,000 like how I put over here inside my session state as we go ahead. 143 00:08:24,000 --> 00:08:24,000 Right. 144 00:08:24,000 --> 00:08:27,000 So right now in this key value pairs I will keep on adding it. 145 00:08:27,000 --> 00:08:32,000 And we will be making sure that we add all these values in our session state. 146 00:08:32,000 --> 00:08:33,000 Okay. 147 00:08:33,000 --> 00:08:36,000 Now, uh, let's go ahead and do one thing. 148 00:08:36,000 --> 00:08:39,000 Let's go ahead and create my prompt. 149 00:08:39,000 --> 00:08:40,000 Okay. 150 00:08:40,000 --> 00:08:45,000 Now prompt is something, uh, you know, by default some text message will be there. 151 00:08:45,000 --> 00:08:54,000 So I'll write from prompt colon equal to okay I'll go ahead and say TX dot chat underscore input okay. 152 00:08:54,000 --> 00:08:57,000 And here I'm going to just go ahead and see chat. 153 00:08:57,000 --> 00:09:02,000 Underscore input is also one very important uh component in Streamlit. 154 00:09:02,000 --> 00:09:05,000 And here we are just going to go ahead and put our placeholder. 155 00:09:05,000 --> 00:09:07,000 So I will go ahead and write my placeholder. 156 00:09:07,000 --> 00:09:10,000 And I'll say hey what is machine learning. 157 00:09:10,000 --> 00:09:14,000 I'm just going to put a simple basic message, okay. 158 00:09:15,000 --> 00:09:17,000 Uh, regarding this chat underscore input. 159 00:09:17,000 --> 00:09:22,000 Now this prompt by default will be having this specific placeholder. 160 00:09:22,000 --> 00:09:24,000 Now I will go inside this. 161 00:09:24,000 --> 00:09:28,000 If this particular prompt is specifically written, I will go ahead and say, hey, let's go ahead and 162 00:09:28,000 --> 00:09:33,000 create this SD dot session state session underscore state. 163 00:09:33,000 --> 00:09:35,000 And I'm going to take this messages. 164 00:09:35,000 --> 00:09:37,000 Um let's see. 165 00:09:37,000 --> 00:09:41,000 Yes messages dot I'm going to append some role. 166 00:09:41,000 --> 00:09:41,000 Right. 167 00:09:41,000 --> 00:09:48,000 And initially whenever I try to put any if I try to ask any query the role, you know what it should 168 00:09:48,000 --> 00:09:49,000 be added. 169 00:09:49,000 --> 00:09:50,000 It should be added as a user. 170 00:09:50,000 --> 00:09:52,000 So this will be my role. 171 00:09:52,000 --> 00:09:52,000 Okay. 172 00:09:52,000 --> 00:09:57,000 And the content that I'm actually going to add is nothing but my prompt. 173 00:09:57,000 --> 00:09:57,000 Okay. 174 00:09:57,000 --> 00:10:01,000 So here I'm going to probably go ahead and create my prompt. 175 00:10:01,000 --> 00:10:02,000 Perfect. 176 00:10:04,000 --> 00:10:11,000 So once we add this then the next step will be that I will go ahead and say TT dot chat underscore message. 177 00:10:12,000 --> 00:10:15,000 And here I'm going to basically use user. 178 00:10:15,000 --> 00:10:22,000 And we are just going to write this prompt dot right of prompt. 179 00:10:22,000 --> 00:10:23,000 Okay. 180 00:10:23,000 --> 00:10:28,000 So what we are basically doing is that whenever we write any prompt, it is going to get appended in 181 00:10:28,000 --> 00:10:30,000 this particular message so that we will be able to read this. 182 00:10:30,000 --> 00:10:30,000 And. 183 00:10:30,000 --> 00:10:31,000 All right. 184 00:10:31,000 --> 00:10:33,000 Uh, so this is also done. 185 00:10:33,000 --> 00:10:38,000 Uh, now the next step will be that, uh, inside this right inside this prompt itself, we will go 186 00:10:38,000 --> 00:10:40,000 ahead and create our LM. 187 00:10:40,000 --> 00:10:42,000 So I'll write chat grok. 188 00:10:42,000 --> 00:10:42,000 And. 189 00:10:45,000 --> 00:10:51,000 I will go ahead and use chat grok underscore API underscore key okay. 190 00:10:51,000 --> 00:10:56,000 Which will be equal to I think I've written right chat grok. 191 00:10:56,000 --> 00:10:59,000 And here have I imported chat grok or not. 192 00:10:59,000 --> 00:11:01,000 Yes I have imported Chat Grok. 193 00:11:01,000 --> 00:11:04,000 So here I will be using my grok API key. 194 00:11:04,000 --> 00:11:04,000 So let's see. 195 00:11:05,000 --> 00:11:07,000 The grok API key has been imported or not. 196 00:11:07,000 --> 00:11:08,000 It's okay. 197 00:11:08,000 --> 00:11:14,000 So we are going to get this grok API key from our uh parameters right. 198 00:11:14,000 --> 00:11:16,000 Which we will be getting it through from the input itself. 199 00:11:16,000 --> 00:11:17,000 Right. 200 00:11:17,000 --> 00:11:23,000 So here what I will do, I will just use this API key that I'm actually going to use it over here okay. 201 00:11:23,000 --> 00:11:26,000 So once this is done, uh, this will basically be my model. 202 00:11:26,000 --> 00:11:29,000 And here I'm just going to use my model underscore name. 203 00:11:29,000 --> 00:11:35,000 You can use any model for now I'm using llama three 8 billion parameter. 204 00:11:35,000 --> 00:11:37,000 And this will be with 8912 okay. 205 00:11:37,000 --> 00:11:38,000 So this will be my model. 206 00:11:38,000 --> 00:11:42,000 And here I'm also going to keep my parameter as streaming is equal to true. 207 00:11:42,000 --> 00:11:44,000 So let me just go ahead and write it over here. 208 00:11:44,000 --> 00:11:47,000 And streaming will be equal to true okay. 209 00:11:47,000 --> 00:11:47,000 Okay. 210 00:11:49,000 --> 00:11:53,000 So this basically becomes my LM model altogether. 211 00:11:53,000 --> 00:11:59,000 Um, I have initialized all the all the tools, whichever I really want to use it. 212 00:11:59,000 --> 00:12:00,000 Okay. 213 00:12:00,000 --> 00:12:02,000 And these are the tools that I initialized. 214 00:12:02,000 --> 00:12:03,000 Right. 215 00:12:03,000 --> 00:12:07,000 Uh, one is uh, the wiki, one is RCF and one is search. 216 00:12:07,000 --> 00:12:10,000 So I will go ahead and create a variable called as tools. 217 00:12:11,000 --> 00:12:14,000 And I will initialize all these tools over here. 218 00:12:14,000 --> 00:12:14,000 Okay. 219 00:12:14,000 --> 00:12:16,000 So let's say I will go ahead and write. 220 00:12:17,000 --> 00:12:19,000 One is search. 221 00:12:19,000 --> 00:12:20,000 Then one is are safe. 222 00:12:20,000 --> 00:12:21,000 Then one is wiki. 223 00:12:22,000 --> 00:12:22,000 Okay. 224 00:12:22,000 --> 00:12:24,000 All these tools is specifically been added. 225 00:12:24,000 --> 00:12:30,000 Now I need to convert this entire tools into an agent so that I will be able to, you know, invoke 226 00:12:30,000 --> 00:12:31,000 this specific agent. 227 00:12:31,000 --> 00:12:33,000 So I will say, hey, this is my search agent. 228 00:12:33,000 --> 00:12:38,000 And here I'm going to basically use this initialize agent function. 229 00:12:38,000 --> 00:12:43,000 And along with this I will be using my uh my tools will be going over here. 230 00:12:43,000 --> 00:12:44,000 So this will be my tools. 231 00:12:44,000 --> 00:12:49,000 This tools will be in the form of list that you have over here. 232 00:12:49,000 --> 00:12:53,000 So all the search search Ar7 wiki we are going to specifically use over here. 233 00:12:53,000 --> 00:12:58,000 And along with this the second parameter will be my LM which I'm actually going to use. 234 00:12:58,000 --> 00:13:01,000 The third will be something called as agent. 235 00:13:01,000 --> 00:13:06,000 So the agent that we are going to basically initialize over here is my agent type dot. 236 00:13:06,000 --> 00:13:10,000 There will be something called as zero short react description. 237 00:13:10,000 --> 00:13:14,000 And there is also something called as chat zero short react description. 238 00:13:14,000 --> 00:13:17,000 And there is one more something called as structured chat zero short react description. 239 00:13:17,000 --> 00:13:22,000 You can definitely see the documentation what exactly this means, but let me give you a brief idea. 240 00:13:22,000 --> 00:13:28,000 What is the main difference between zero shot react description and zero shot react description? 241 00:13:28,000 --> 00:13:36,000 It basically lies in how they handle our context and memory, how they generate prompts for the language 242 00:13:36,000 --> 00:13:36,000 model. 243 00:13:36,000 --> 00:13:42,000 See in the case of zero shot, zero shot, uh, react description, they don't rely on the chat history, 244 00:13:42,000 --> 00:13:43,000 right? 245 00:13:43,000 --> 00:13:48,000 Uh, it makes a decision based on the current input only without considering any previous action. 246 00:13:48,000 --> 00:13:54,000 Similarly, on the other hand, if I use Chad zero shot react description, it uses a chat history to 247 00:13:54,000 --> 00:13:59,000 remember the context of the chat and the history of the conversation, and it accepts a certain structure 248 00:13:59,000 --> 00:14:02,000 in the chat history and might raise an error if it does not find it. 249 00:14:02,000 --> 00:14:08,000 Okay, so in our case, I will just go ahead and use this zero shot react description. 250 00:14:08,000 --> 00:14:14,000 And here I'm also going to make sure that I give one more additional parameter which is called as handling 251 00:14:14,000 --> 00:14:14,000 pass. 252 00:14:14,000 --> 00:14:21,000 So here I'm going to go ahead and write handling underscore pass is passing. 253 00:14:21,000 --> 00:14:21,000 Sorry. 254 00:14:21,000 --> 00:14:24,000 Passing underscore errors is equal to true. 255 00:14:24,000 --> 00:14:29,000 I'm also saying that if you face any errors please make sure that you also pass those errors okay. 256 00:14:29,000 --> 00:14:34,000 So this are the basic parameters that I specifically require over here. 257 00:14:34,000 --> 00:14:34,000 Right. 258 00:14:34,000 --> 00:14:39,000 So here inside all these things I've actually given this Let me make the spelling right. 259 00:14:39,000 --> 00:14:40,000 Search underscore agent. 260 00:14:40,000 --> 00:14:41,000 Okay. 261 00:14:41,000 --> 00:14:43,000 And this is how we basically create this agent. 262 00:14:44,000 --> 00:14:48,000 Now it's time that we go ahead and have this kind of conversation. 263 00:14:48,000 --> 00:14:49,000 And it should be able to give us an answer. 264 00:14:49,000 --> 00:14:54,000 So I'll try it with s t dot chat underscore message. 265 00:14:54,000 --> 00:14:59,000 And I'm going to give the history over here or sorry message of assistant. 266 00:14:59,000 --> 00:15:00,000 Right. 267 00:15:00,000 --> 00:15:02,000 So I'll go ahead and give like assistant. 268 00:15:02,000 --> 00:15:03,000 Okay. 269 00:15:03,000 --> 00:15:06,000 Whenever the assistant is giving me any kind of response. 270 00:15:06,000 --> 00:15:11,000 So what I'm actually going to do is that I'm also going to call this Streamlit callback handler. 271 00:15:11,000 --> 00:15:16,000 I know there are multiple things that I'm actually showing you, but this is the best way of creating 272 00:15:16,000 --> 00:15:16,000 this, right? 273 00:15:16,000 --> 00:15:22,000 So once I use the Streamlit callback handler, understand what it is doing it is displaying. 274 00:15:22,000 --> 00:15:27,000 It is used to display the thoughts and action of an agent in an interactive Streamlit app. 275 00:15:27,000 --> 00:15:32,000 So now whenever my agent is basically communicating with itself, right? 276 00:15:32,000 --> 00:15:33,000 It should be. 277 00:15:33,000 --> 00:15:37,000 We should be able to see that and that will be able to see once I will execute it. 278 00:15:37,000 --> 00:15:40,000 So here I will be using an SD dot container. 279 00:15:40,000 --> 00:15:42,000 So container over here. 280 00:15:42,000 --> 00:15:47,000 And if you see what exactly this container is it in turn insert a multi element container. 281 00:15:47,000 --> 00:15:52,000 Inserts an invisible container into your app that can be used to hold multiple elements. 282 00:15:52,000 --> 00:15:55,000 This allows you to, for example, insert multiple elements into your app out of order. 283 00:15:55,000 --> 00:15:57,000 Okay, that is the reason we are specifically using. 284 00:15:57,000 --> 00:16:02,000 You'll be seeing will be able to see that block wise. 285 00:16:02,000 --> 00:16:04,000 You'll be having that entire information. 286 00:16:04,000 --> 00:16:10,000 And then we will go ahead and create one more parameter called as expand underscore new underscore thoughts. 287 00:16:10,000 --> 00:16:14,000 And I'll say, please, whatever thoughts you have show me that detailed explanation. 288 00:16:14,000 --> 00:16:19,000 And if I say expand, right, so every thought will probably just get opened. 289 00:16:19,000 --> 00:16:22,000 We can also open this, but right now I'll just keep it to false. 290 00:16:22,000 --> 00:16:29,000 So again this expand new thoughts you'll be able to see once I probably, uh, run my entire application. 291 00:16:29,000 --> 00:16:30,000 You'll be seeing that by default it will be closed. 292 00:16:30,000 --> 00:16:32,000 You can also open it okay. 293 00:16:32,000 --> 00:16:36,000 Now let's go ahead and get the response so quickly I will go ahead and write. 294 00:16:36,000 --> 00:16:42,000 Response is equal to search underscore agent dot run. 295 00:16:42,000 --> 00:16:51,000 I will run this and while running I will also give all my session state dot messages okay. 296 00:16:51,000 --> 00:16:54,000 And I will say hey you can also make a callback. 297 00:16:54,000 --> 00:16:59,000 So there is a parameter which is called as callbacks with whatever callback we have actually created, 298 00:16:59,000 --> 00:16:59,000 right? 299 00:16:59,000 --> 00:17:01,000 Streamlit callback handler. 300 00:17:01,000 --> 00:17:08,000 So I'll go ahead and say TX comma cb okay, once I do this now the next step will be that I will go 301 00:17:08,000 --> 00:17:15,000 ahead and again append it's t dot session, underscore state dot messages dot append. 302 00:17:15,000 --> 00:17:21,000 And here I'm going to specifically use my role as assistant because this will be given by the assistant 303 00:17:21,000 --> 00:17:23,000 itself okay. 304 00:17:23,000 --> 00:17:30,000 And my whatever response I'm actually specifically getting it will be inserted into the content part. 305 00:17:30,000 --> 00:17:33,000 So here I have this response okay. 306 00:17:33,000 --> 00:17:39,000 Now quickly let's go ahead and write SD dot write response okay. 307 00:17:39,000 --> 00:17:42,000 So this is what it is. 308 00:17:42,000 --> 00:17:45,000 And uh I think we have executed this entire thing. 309 00:17:45,000 --> 00:17:48,000 Uh, let's execute this entire thing and see whether everything is working fine or not. 310 00:17:48,000 --> 00:17:50,000 But I think it should work. 311 00:17:50,000 --> 00:17:53,000 Just to give you a quick revision, we have created all the tools wrapper. 312 00:17:53,000 --> 00:17:57,000 Then we use this tools parameter over here which will be included. 313 00:17:57,000 --> 00:18:04,000 We created our search agent LM agent type this, this this and we are good to go with respect to this 314 00:18:04,000 --> 00:18:04,000 okay. 315 00:18:04,000 --> 00:18:08,000 And this is completely with the open source uh LM models. 316 00:18:08,000 --> 00:18:14,000 Now I will quickly go ahead and write CD which folder it is in five search engine. 317 00:18:14,000 --> 00:18:17,000 So I will go ahead and write right five tab search engine. 318 00:18:17,000 --> 00:18:19,000 Now quickly let's go ahead and run it. 319 00:18:19,000 --> 00:18:21,000 Streamlit run app.py okay. 320 00:18:21,000 --> 00:18:24,000 So once I run this I hope so. 321 00:18:24,000 --> 00:18:26,000 It should get executed. 322 00:18:26,000 --> 00:18:29,000 Okay, so here it is. 323 00:18:29,000 --> 00:18:32,000 Um, HDR object has no attribute. 324 00:18:32,000 --> 00:18:33,000 Let's see. 325 00:18:33,000 --> 00:18:36,000 So it says on line 36 I'm having this particular error. 326 00:18:36,000 --> 00:18:37,000 Let's see okay. 327 00:18:37,000 --> 00:18:41,000 Line 36 I am I'm having this particular error. 328 00:18:41,000 --> 00:18:45,000 Let me go ahead and check to this. 329 00:18:45,000 --> 00:18:47,000 What is exactly the problem over here. 330 00:18:47,000 --> 00:18:55,000 So if I probably see this message 36 it is chat underscore message message dot okay I made one specific 331 00:18:55,000 --> 00:18:56,000 simple mistake. 332 00:18:56,000 --> 00:18:58,000 I should have close the brackets over here. 333 00:18:59,000 --> 00:19:01,000 And no need of this specific brackets. 334 00:19:02,000 --> 00:19:08,000 So whenever you get an error, don't feel afraid of the things that you are seeing. 335 00:19:08,000 --> 00:19:10,000 Instead, just try to fix it. 336 00:19:10,000 --> 00:19:12,000 Now this looks absolutely fabulous, right? 337 00:19:13,000 --> 00:19:17,000 Uh, quickly, let me just go ahead and take my grok key. 338 00:19:17,000 --> 00:19:21,000 I will copy this entire thing over here. 339 00:19:21,000 --> 00:19:26,000 I'll go over here and press enter, and let's go ahead and do. 340 00:19:26,000 --> 00:19:26,000 Okay. 341 00:19:26,000 --> 00:19:28,000 I did not write one more condition over here. 342 00:19:28,000 --> 00:19:28,000 Okay. 343 00:19:29,000 --> 00:19:32,000 Um, and that is the reason it may give us an error. 344 00:19:32,000 --> 00:19:41,000 I'll say if prompt and and write and the API key, I have to probably give this API key. 345 00:19:41,000 --> 00:19:42,000 Okay. 346 00:19:42,000 --> 00:19:43,000 Now I think it should run. 347 00:19:44,000 --> 00:19:46,000 Oh let's go ahead and copy and paste it. 348 00:19:46,000 --> 00:19:47,000 I will press enter. 349 00:19:48,000 --> 00:19:54,000 Now I will go ahead and write what is generative AI? 350 00:19:56,000 --> 00:19:58,000 Oh God, it is getting displayed. 351 00:19:58,000 --> 00:20:00,000 Why it is getting displayed, we'll see to it. 352 00:20:00,000 --> 00:20:04,000 But at least the functionality is working over here, right? 353 00:20:04,000 --> 00:20:05,000 So this is my key. 354 00:20:05,000 --> 00:20:06,000 The key should not get displayed over here. 355 00:20:06,000 --> 00:20:08,000 But it got displayed. 356 00:20:08,000 --> 00:20:08,000 Okay. 357 00:20:08,000 --> 00:20:09,000 Let's see. 358 00:20:09,000 --> 00:20:13,000 Uh, am I printing somewhere the API key? 359 00:20:13,000 --> 00:20:15,000 So API key is over here. 360 00:20:16,000 --> 00:20:16,000 Okay. 361 00:20:16,000 --> 00:20:20,000 It is also getting appended in this specific message okay. 362 00:20:20,000 --> 00:20:22,000 But let's see whether everything is working fine or not. 363 00:20:22,000 --> 00:20:25,000 So here you can see I asked what is machine learning okay. 364 00:20:26,000 --> 00:20:29,000 Um or here you can see what can I search. 365 00:20:29,000 --> 00:20:30,000 Okay. 366 00:20:30,000 --> 00:20:33,000 Let's see I'll ask what is machine learning. 367 00:20:33,000 --> 00:20:35,000 Let's hit one more time okay. 368 00:20:35,000 --> 00:20:37,000 So again this is getting displayed. 369 00:20:37,000 --> 00:20:41,000 How can I help you search okay. 370 00:20:41,000 --> 00:20:42,000 There there is some problem over here. 371 00:20:42,000 --> 00:20:44,000 Let's see some problem. 372 00:20:44,000 --> 00:20:45,000 Some problem. 373 00:20:45,000 --> 00:20:51,000 So whenever I'm hitting a query write prompt, then it is writing the specific prompt. 374 00:20:51,000 --> 00:20:55,000 The simple reason is that because it is considering this as a prompt. 375 00:20:55,000 --> 00:20:56,000 Okay. 376 00:20:56,000 --> 00:21:02,000 So uh, that is the reason you will be able to see that when I'm hitting this and I'm putting this particular 377 00:21:02,000 --> 00:21:03,000 condition. 378 00:21:03,000 --> 00:21:04,000 Okay, let's do one thing. 379 00:21:04,000 --> 00:21:05,000 Let's remove this for now. 380 00:21:05,000 --> 00:21:06,000 Onwards. 381 00:21:06,000 --> 00:21:06,000 Okay. 382 00:21:06,000 --> 00:21:08,000 And let's reload it. 383 00:21:09,000 --> 00:21:10,000 We'll go ahead and reload it. 384 00:21:10,000 --> 00:21:13,000 I will quickly do one thing. 385 00:21:13,000 --> 00:21:15,000 I will put my API key, I'll press enter. 386 00:21:16,000 --> 00:21:20,000 And now I will just go ahead and write what is machine learning. 387 00:21:21,000 --> 00:21:27,000 So now I'm getting the input because the API key was basically getting uh, you know hidden over there. 388 00:21:27,000 --> 00:21:32,000 So now you see, see see the perfect thing As soon as I searched what is machine learning? 389 00:21:32,000 --> 00:21:34,000 It it went to directly to Wikipedia. 390 00:21:34,000 --> 00:21:38,000 Now, see, the interaction is also basically happening right here. 391 00:21:38,000 --> 00:21:41,000 You can see automatically Wikipedia is able to give some more information. 392 00:21:41,000 --> 00:21:42,000 Okay. 393 00:21:42,000 --> 00:21:45,000 Then it automatically took the examples. 394 00:21:45,000 --> 00:21:47,000 Then Wikipedia is saying hey I have more details. 395 00:21:47,000 --> 00:21:48,000 Go ahead and see this. 396 00:21:48,000 --> 00:21:48,000 Right. 397 00:21:49,000 --> 00:21:51,000 Uh, and here you can see, hey, Wikipedia is also there. 398 00:21:51,000 --> 00:21:51,000 Right. 399 00:21:51,000 --> 00:21:55,000 And all the information is here is the response. 400 00:21:55,000 --> 00:21:56,000 All the information is basically coming up. 401 00:21:56,000 --> 00:21:57,000 Right. 402 00:21:57,000 --> 00:21:59,000 And finally this is the output that I'm able to get. 403 00:21:59,000 --> 00:22:02,000 See, so much interaction is basically happening right. 404 00:22:03,000 --> 00:22:06,000 Uh this content is basically getting displayed I'm printing it right. 405 00:22:06,000 --> 00:22:07,000 Action. 406 00:22:07,000 --> 00:22:09,000 Everything is going on okay okay. 407 00:22:09,000 --> 00:22:13,000 Tell me about the research paper. 408 00:22:13,000 --> 00:22:14,000 Okay. 409 00:22:15,000 --> 00:22:18,000 Um, attention is all you need. 410 00:22:19,000 --> 00:22:20,000 and I'll go ahead and press enter. 411 00:22:21,000 --> 00:22:23,000 See, it went to arXiv. 412 00:22:23,000 --> 00:22:24,000 Amazing. 413 00:22:24,000 --> 00:22:24,000 Right. 414 00:22:24,000 --> 00:22:27,000 So here now it is thinking now Wikipedia came into picture. 415 00:22:27,000 --> 00:22:28,000 Okay. 416 00:22:29,000 --> 00:22:31,000 Then suddenly it went to search. 417 00:22:31,000 --> 00:22:33,000 Attention is all you need arXiv paper. 418 00:22:33,000 --> 00:22:34,000 Then again it went to search. 419 00:22:34,000 --> 00:22:37,000 See how the interaction is basically happening, right? 420 00:22:38,000 --> 00:22:43,000 I thought I understood that the user wants to know more about the research paper right over here. 421 00:22:43,000 --> 00:22:44,000 This is the thought. 422 00:22:44,000 --> 00:22:45,000 Here is the continuation. 423 00:22:45,000 --> 00:22:48,000 The Wikipedia search found a relevant way, but it is not a research paper. 424 00:22:48,000 --> 00:22:50,000 See, attention is all you need. 425 00:22:50,000 --> 00:22:51,000 You don't want to know more about this. 426 00:22:51,000 --> 00:22:54,000 Or I say attention is no good or save result was found. 427 00:22:55,000 --> 00:22:56,000 Amazing right? 428 00:22:56,000 --> 00:23:01,000 So this automatically this is like you know you have independent tools who are interacting with the 429 00:23:01,000 --> 00:23:02,000 outside world. 430 00:23:02,000 --> 00:23:07,000 It's like I'm having four people who's probably doing all the search and who is probably, you know, 431 00:23:07,000 --> 00:23:09,000 making sure that all the things basically happen. 432 00:23:09,000 --> 00:23:13,000 And one error is over here handling parsing errors. 433 00:23:13,000 --> 00:23:14,000 Okay. 434 00:23:14,000 --> 00:23:19,000 An output parser, this is error could not pass LM thought I have found multiple research paper okay. 435 00:23:19,000 --> 00:23:20,000 So let's see. 436 00:23:20,000 --> 00:23:21,000 We'll try to fix this also. 437 00:23:21,000 --> 00:23:27,000 But uh uh I think uh, this is just because uh, they have actually got multiple results, like no one 438 00:23:27,000 --> 00:23:30,000 is coming to the right purpose right over here. 439 00:23:30,000 --> 00:23:31,000 So now let's see. 440 00:23:31,000 --> 00:23:32,000 I'll go ahead and write. 441 00:23:32,000 --> 00:23:34,000 Attention is all you need. 442 00:23:34,000 --> 00:23:34,000 Okay. 443 00:23:34,000 --> 00:23:37,000 And I will just go and search for this research paper. 444 00:23:38,000 --> 00:23:39,000 Okay. 445 00:23:40,000 --> 00:23:41,000 I will say hey. 446 00:23:44,000 --> 00:23:47,000 Provide a summary of. 447 00:23:50,000 --> 00:23:55,000 So here you can see that automatically all this summary will be found out by ourself. 448 00:23:56,000 --> 00:24:01,000 Right here all the information once a summary of attention is all you need. 449 00:24:01,000 --> 00:24:02,000 Then again it is doing this. 450 00:24:02,000 --> 00:24:06,000 It did not have to probably go with or ask any other tools. 451 00:24:06,000 --> 00:24:08,000 The user is asking title this, this this. 452 00:24:08,000 --> 00:24:10,000 All the information is over here. 453 00:24:11,000 --> 00:24:16,000 The user is asking for the summary of the research paper, which is a published paper on arXiv. 454 00:24:18,000 --> 00:24:19,000 Right. 455 00:24:19,000 --> 00:24:22,000 So this is going on and this interaction will keep on going ahead. 456 00:24:22,000 --> 00:24:24,000 And finally you'll be able to see some output. 457 00:24:24,000 --> 00:24:27,000 It will be able to probably provide you okay. 458 00:24:28,000 --> 00:24:35,000 But uh, here I see that it is also going into that entire loop I agent stopped due to iteration limit. 459 00:24:35,000 --> 00:24:36,000 Yes. 460 00:24:36,000 --> 00:24:39,000 So this can also be one type of problem. 461 00:24:39,000 --> 00:24:43,000 Uh, we'll see in the upcoming videos how we can fix this kind of problem also. 462 00:24:43,000 --> 00:24:48,000 Okay, so let me just go ahead and ask, just like this okay, I want. 463 00:24:50,000 --> 00:24:55,000 Tell me about the research paper. 464 00:24:56,000 --> 00:24:57,000 Nothing but this. 465 00:24:58,000 --> 00:25:05,000 So here you can see the user is asking about a specific I have a tool to search for it on arXiv right 466 00:25:05,000 --> 00:25:08,000 then Wikipedia, then all this interaction is coming. 467 00:25:08,000 --> 00:25:10,000 And this paper is this authored by so and so. 468 00:25:10,000 --> 00:25:12,000 All the information is over here. 469 00:25:12,000 --> 00:25:12,000 Right. 470 00:25:12,000 --> 00:25:17,000 Let me just go ahead and search for something like what is generative AI. 471 00:25:17,000 --> 00:25:21,000 Now I think it should probably go with Wikipedia or search. 472 00:25:21,000 --> 00:25:22,000 Right. 473 00:25:22,000 --> 00:25:24,000 So search is duck duck go search right. 474 00:25:24,000 --> 00:25:28,000 Thought the user has multiple ranges ranging from general topics like what is the machine learning to 475 00:25:28,000 --> 00:25:29,000 specifically see. 476 00:25:29,000 --> 00:25:31,000 It is also using the chat history. 477 00:25:31,000 --> 00:25:31,000 Right. 478 00:25:31,000 --> 00:25:33,000 And based on that it is giving you the entire response. 479 00:25:34,000 --> 00:25:34,000 Right? 480 00:25:34,000 --> 00:25:36,000 I've already provided a brief overview. 481 00:25:36,000 --> 00:25:37,000 See, it's saying that again. 482 00:25:37,000 --> 00:25:38,000 I'll not go ahead and provide it. 483 00:25:38,000 --> 00:25:40,000 I've already provided it right. 484 00:25:40,000 --> 00:25:41,000 Generative AI right. 485 00:25:41,000 --> 00:25:43,000 And here are all the answers. 486 00:25:43,000 --> 00:25:46,000 So I hope you are able to understand this. 487 00:25:46,000 --> 00:25:51,000 Uh, again, um, if you want to go ahead and explore more examples, you can go ahead and check it 488 00:25:51,000 --> 00:25:51,000 out over here. 489 00:25:51,000 --> 00:25:52,000 Right. 490 00:25:52,000 --> 00:25:55,000 So yes, this was it about this entire end to end projects. 491 00:25:55,000 --> 00:25:57,000 Uh, we'll see, uh, some more errors that we are getting. 492 00:25:57,000 --> 00:26:01,000 I think I need to have a look why we got that specific error, but I don't think so. 493 00:26:01,000 --> 00:26:04,000 It is a major issue with respect to that. 494 00:26:04,000 --> 00:26:08,000 Unless and until we are having a good interaction between the agents, I think we can work out. 495 00:26:08,000 --> 00:26:14,000 One assignment that I really want to give you is that try to integrate a PDF document Q&A along with 496 00:26:14,000 --> 00:26:18,000 these tools, and try to add it also and try to see that how the interaction will basically happen. 497 00:26:18,000 --> 00:26:20,000 So yes, this was it for my side. 498 00:26:20,000 --> 00:26:21,000 I'll see you all in the next video. 499 00:26:21,000 --> 00:26:21,000 Thank you.