1 00:00:00,000 --> 00:00:01,000 Hello guys. 2 00:00:01,000 --> 00:00:04,000 So we are going to continue the discussion with respect to the embedding techniques. 3 00:00:04,000 --> 00:00:11,000 In our previous video, we have already seen how we can use OpenAI embeddings and uh, convert your 4 00:00:11,000 --> 00:00:13,000 text or document into vectors. 5 00:00:13,000 --> 00:00:16,000 Let's say you do not want to use OpenAI API. 6 00:00:16,000 --> 00:00:19,000 Then you can also go ahead with open source models. 7 00:00:19,000 --> 00:00:25,000 And one of the way of using this open source models, uh, and use the embedding technique, is through 8 00:00:25,000 --> 00:00:27,000 this platform, which is called as Lama. 9 00:00:27,000 --> 00:00:32,000 Now in this video I'll show you how you can probably go ahead and do the setup of this olama and how 10 00:00:32,000 --> 00:00:33,000 you can run the code. 11 00:00:33,000 --> 00:00:36,000 Okay, so what exactly is Olama? 12 00:00:36,000 --> 00:00:42,000 Olama is something is a kind of platform where you will be able to use open source LLM models like llama 13 00:00:42,000 --> 00:00:44,000 353, Mistral and Gamma. 14 00:00:44,000 --> 00:00:48,000 These are various open source models that are available right now. 15 00:00:48,000 --> 00:00:54,000 And you can probably run all these models in your local machine itself, in your local workstation. 16 00:00:54,000 --> 00:00:55,000 Right. 17 00:00:55,000 --> 00:01:00,000 So here you can see get up and start running with large language models. 18 00:01:00,000 --> 00:01:04,000 I'll be showing you, uh, this particular platform. 19 00:01:04,000 --> 00:01:08,000 You can download it in your local machine and then you can download any of this specific models. 20 00:01:08,000 --> 00:01:10,000 And then you can use this LM models. 21 00:01:10,000 --> 00:01:10,000 Okay. 22 00:01:11,000 --> 00:01:17,000 Uh, now when we discuss about embedding techniques with respect to this llama 353 Mistral gamma. 23 00:01:17,000 --> 00:01:21,000 There are also different different embedding techniques with respect to that, right. 24 00:01:21,000 --> 00:01:26,000 If I go ahead and use llama embedding, then let's say that I have installed llama two or llama three. 25 00:01:26,000 --> 00:01:30,000 Um, then I'll be able to use the embedding techniques that are available over here. 26 00:01:30,000 --> 00:01:32,000 So let me just go ahead step by step. 27 00:01:32,000 --> 00:01:36,000 First of all, we'll see how to download this particular llama platform. 28 00:01:36,000 --> 00:01:38,000 Uh I will go ahead and click on this llama. 29 00:01:38,000 --> 00:01:42,000 First of all you need to go to this particular website called as alumni.com. 30 00:01:42,000 --> 00:01:46,000 And if I go to this GitHub of this, it is a complete open source. 31 00:01:46,000 --> 00:01:50,000 Here you will be able to see that it supports all these open source model okay. 32 00:01:50,000 --> 00:01:54,000 So here you can see all these open source models like llama three 8 billion parameters. 33 00:01:54,000 --> 00:01:56,000 Llama 370 billion parameters. 34 00:01:57,000 --> 00:01:59,000 Phi three mini 3.8 billion parameters. 35 00:01:59,000 --> 00:02:00,000 Then you have gamma. 36 00:02:00,000 --> 00:02:03,000 Google gamma 2,000,000,007 billion parameters. 37 00:02:03,000 --> 00:02:05,000 Then you have Mistral moon dream. 38 00:02:05,000 --> 00:02:07,000 neural chart, star link code llama. 39 00:02:07,000 --> 00:02:10,000 So we will be using all these specific models as we go ahead. 40 00:02:11,000 --> 00:02:14,000 We'll be creating end to end projects by using all these models okay. 41 00:02:14,000 --> 00:02:17,000 So first of all I will go ahead and click on download. 42 00:02:17,000 --> 00:02:18,000 As soon as I go ahead and click on download. 43 00:02:18,000 --> 00:02:24,000 You have this particular platform in all the operating system like uh, Mac OS, then you have Linux, 44 00:02:24,000 --> 00:02:26,000 then you have windows, right? 45 00:02:26,000 --> 00:02:32,000 So uh, right now, uh, since I have a windows machine, I will go ahead and download this particular 46 00:02:32,000 --> 00:02:33,000 windows exe file. 47 00:02:33,000 --> 00:02:33,000 Okay. 48 00:02:33,000 --> 00:02:38,000 For this windows we will have uh exe file that will get downloaded. 49 00:02:38,000 --> 00:02:41,000 Uh, but one condition is that you really need to have Windows 10 or later. 50 00:02:41,000 --> 00:02:43,000 Then only this will work okay. 51 00:02:43,000 --> 00:02:45,000 So this is one of the criteria. 52 00:02:45,000 --> 00:02:48,000 So in order to download it just go ahead and click on this. 53 00:02:48,000 --> 00:02:53,000 So as soon as you click on this you'll be able to see that I will be getting an EXE file that gets installed. 54 00:02:53,000 --> 00:02:53,000 Okay. 55 00:02:54,000 --> 00:02:58,000 Uh, once this x file gets installed, the installation is pretty much simple and easy. 56 00:02:58,000 --> 00:03:03,000 So here what I will do is that I'll just go ahead and double click that particular x file once this 57 00:03:03,000 --> 00:03:04,000 gets downloaded. 58 00:03:04,000 --> 00:03:07,000 And just keep on clicking next okay. 59 00:03:07,000 --> 00:03:13,000 Once you finish the installation here in the bottom section you'll be seeing one icon. 60 00:03:13,000 --> 00:03:16,000 So this kind of icon you'll be able to see it. 61 00:03:16,000 --> 00:03:21,000 Uh, I'm not able to zoom in much more over here, but just here you can see this particular icon will 62 00:03:21,000 --> 00:03:22,000 get enabled. 63 00:03:23,000 --> 00:03:26,000 Now this icon is nothing, but it is the icon of Allama. 64 00:03:26,000 --> 00:03:32,000 Okay, so, uh, first of all, you just download the exe file, double click on that and keep on just 65 00:03:32,000 --> 00:03:35,000 pressing next, next next and do the installation. 66 00:03:35,000 --> 00:03:38,000 As soon as you do, the installation icon will be running in the back end. 67 00:03:38,000 --> 00:03:41,000 Okay, now let's go back. 68 00:03:41,000 --> 00:03:44,000 And after the installation I've already done the installation. 69 00:03:44,000 --> 00:03:48,000 So I will show you how you can go ahead and download this kind of models. 70 00:03:48,000 --> 00:03:48,000 Okay. 71 00:03:49,000 --> 00:03:53,000 So after you do the installation just go ahead and open your command prompt. 72 00:03:53,000 --> 00:03:59,000 Now in this prompt command prompt, let's say I want to go ahead and use llama three or llama two or 73 00:03:59,000 --> 00:04:00,000 gamma model okay. 74 00:04:00,000 --> 00:04:03,000 So I will just go ahead and use this or run this command. 75 00:04:03,000 --> 00:04:03,000 Right. 76 00:04:03,000 --> 00:04:09,000 So let's say if I go ahead and write all of my run gamma 2 billion parameters, I will go over here, 77 00:04:09,000 --> 00:04:10,000 press paste it. 78 00:04:10,000 --> 00:04:19,000 So allow me to run run gamma 2 billion parameter as soon as I run this command. 79 00:04:19,000 --> 00:04:24,000 If this particular model is not downloaded in my local right. 80 00:04:24,000 --> 00:04:31,000 So first of all, what it will do, it will pull this entire gamma two B model, uh, and then it will 81 00:04:31,000 --> 00:04:32,000 download it in my local machine. 82 00:04:32,000 --> 00:04:38,000 First of all, let's say if it is already downloaded then this chatbot will automatically get started. 83 00:04:38,000 --> 00:04:39,000 So let me go ahead and press enter. 84 00:04:40,000 --> 00:04:44,000 So here you'll be able to see that I have already downloaded this model. 85 00:04:44,000 --> 00:04:45,000 Okay. 86 00:04:45,000 --> 00:04:50,000 Uh, so here you'll be seeing hey nothing has got downloaded because already in my local machine it 87 00:04:50,000 --> 00:04:51,000 has got downloaded. 88 00:04:51,000 --> 00:04:55,000 If you're doing it for the first time initially, this particular model needs to get downloaded. 89 00:04:55,000 --> 00:05:01,000 So as soon as you write this command or run gamma to be so, it will keep on downloading and it will 90 00:05:01,000 --> 00:05:05,000 take some amount of time because this particular file is 1.4 GB. 91 00:05:05,000 --> 00:05:05,000 Okay. 92 00:05:05,000 --> 00:05:08,000 So once you have this you can go ahead and chat it. 93 00:05:08,000 --> 00:05:08,000 Hey. 94 00:05:08,000 --> 00:05:09,000 Hi. 95 00:05:09,000 --> 00:05:10,000 Who are you? 96 00:05:11,000 --> 00:05:11,000 Okay. 97 00:05:11,000 --> 00:05:13,000 I'm a large language model trained by Google. 98 00:05:13,000 --> 00:05:17,000 I'm capable of engaging in a whole wide range of conversation on various topics. 99 00:05:17,000 --> 00:05:21,000 Okay, so here you have all the information you can ask any question as such. 100 00:05:21,000 --> 00:05:25,000 Similarly, you can go ahead and play with any model that you want, any model that you want to work 101 00:05:25,000 --> 00:05:30,000 with like a llama three llama 370 billion 5353 is from Microsoft, right? 102 00:05:30,000 --> 00:05:36,000 Any of the models that you really want to play now, once you do this installation and everything is 103 00:05:36,000 --> 00:05:41,000 working fine from the command prompt, now it's time that we go ahead and start our coding okay. 104 00:05:41,000 --> 00:05:46,000 So so first of all, let me just go ahead and open my, uh, Allama embedding. 105 00:05:46,000 --> 00:05:50,000 So here, uh, what I will do is that I will just show, hide my face. 106 00:05:50,000 --> 00:05:51,000 Okay. 107 00:05:51,000 --> 00:05:58,000 So here you can see Allama supports embedding models, making it possible to build a retrieval augmented 108 00:05:58,000 --> 00:05:58,000 generation. 109 00:05:58,000 --> 00:06:01,000 So this is basically a Rag application. 110 00:06:01,000 --> 00:06:04,000 The based on the architecture that I have actually discussed okay. 111 00:06:04,000 --> 00:06:08,000 Now first of all one of the library that you require is lang chain underscore community. 112 00:06:08,000 --> 00:06:09,000 Okay. 113 00:06:09,000 --> 00:06:12,000 So I'll just go ahead and write lang chain dash community. 114 00:06:12,000 --> 00:06:14,000 Let me, uh, use this particular thing. 115 00:06:14,000 --> 00:06:20,000 And along with this, what I am also going to do is that, uh, I'll talk about one more library that 116 00:06:20,000 --> 00:06:20,000 will be required. 117 00:06:20,000 --> 00:06:22,000 But before this, let's go ahead and install this. 118 00:06:22,000 --> 00:06:23,000 Okay. 119 00:06:23,000 --> 00:06:27,000 So here I'm just going to go ahead and write pip install minus r requirements.txt. 120 00:06:27,000 --> 00:06:31,000 And here you'll be able to see that if it is already installed. 121 00:06:31,000 --> 00:06:35,000 Uh, I've already done the installation so you can see that it has got successfully installed. 122 00:06:35,000 --> 00:06:36,000 Okay. 123 00:06:36,000 --> 00:06:41,000 Now in order to use the embedding, what I am actually going to do is that I'll go ahead and write from 124 00:06:41,000 --> 00:06:42,000 long chain. 125 00:06:43,000 --> 00:06:44,000 From long chain. 126 00:06:44,000 --> 00:06:47,000 Uh, before that let me just go ahead and select my kernel. 127 00:06:47,000 --> 00:06:47,000 Okay. 128 00:06:47,000 --> 00:06:52,000 Don't forget to do that from long chain underscore community I'm just going to import. 129 00:06:53,000 --> 00:06:54,000 Sorry. 130 00:06:54,000 --> 00:06:57,000 It should be Langston underscore community dot embeddings. 131 00:06:58,000 --> 00:06:58,000 Okay. 132 00:06:58,000 --> 00:07:00,000 Embeddings I'll be importing. 133 00:07:01,000 --> 00:07:01,000 Oh. 134 00:07:01,000 --> 00:07:02,000 Llama embeddings okay. 135 00:07:02,000 --> 00:07:05,000 So this is the embedding technique that we are going to use. 136 00:07:05,000 --> 00:07:09,000 So once this gets executed you can see that it has got successfully executed. 137 00:07:09,000 --> 00:07:12,000 Now I'll go ahead and create my embeddings okay. 138 00:07:12,000 --> 00:07:14,000 Like how we did it with OpenAI. 139 00:07:14,000 --> 00:07:18,000 And here I'm going to give my llama my embedding. 140 00:07:18,000 --> 00:07:18,000 Okay. 141 00:07:18,000 --> 00:07:21,000 I'm initializing my all of my embedding. 142 00:07:21,000 --> 00:07:27,000 Remember, once we execute this by default, uh, this all of my embeddings for which LM model it will 143 00:07:27,000 --> 00:07:31,000 be probably taking up, uh, it will be taking up for the llama two model. 144 00:07:31,000 --> 00:07:33,000 So here I will just go ahead and write a comment. 145 00:07:33,000 --> 00:07:39,000 By default it uses llama two. 146 00:07:39,000 --> 00:07:43,000 Now if you really want to use the llama two so what I will do, I'll just go ahead and open my command 147 00:07:43,000 --> 00:07:44,000 prompt. 148 00:07:44,000 --> 00:07:47,000 Let me just execute this okay. 149 00:07:47,000 --> 00:07:49,000 Control Z you just do control Z and exit. 150 00:07:49,000 --> 00:07:53,000 So if I want to use this by default if I'm giving llama two okay. 151 00:07:54,000 --> 00:07:56,000 Um over here uh, or this is taking llama two. 152 00:07:56,000 --> 00:08:00,000 So we need to first of all install llama two okay. 153 00:08:00,000 --> 00:08:03,000 Without installing llama two, it will not work. 154 00:08:03,000 --> 00:08:03,000 Okay. 155 00:08:03,000 --> 00:08:08,000 So that basically means here in llama embedding I will just go ahead and write. 156 00:08:08,000 --> 00:08:10,000 My model is equal to. 157 00:08:10,000 --> 00:08:12,000 And here I will give my model name. 158 00:08:12,000 --> 00:08:13,000 Right. 159 00:08:13,000 --> 00:08:16,000 Let's say as I said by default it uses llama two. 160 00:08:16,000 --> 00:08:21,000 Since you know that I have already downloaded which model I have downloaded this gamma two B right. 161 00:08:21,000 --> 00:08:27,000 So I will just go ahead and copy this particular model and I will paste it over here okay. 162 00:08:27,000 --> 00:08:30,000 So once I go ahead and execute it. 163 00:08:31,000 --> 00:08:34,000 here, you can see that it has got executed successfully. 164 00:08:34,000 --> 00:08:39,000 Now this embedding technique is basically uh see the base URL is running over here. 165 00:08:39,000 --> 00:08:44,000 The model is to be and we are using the embedding technique of this specific model okay. 166 00:08:44,000 --> 00:08:49,000 Now I will just go ahead and use some text okay. 167 00:08:50,000 --> 00:08:56,000 Uh let's say one of my text will be something like this R one is equal to I will write, hey, Allama, 168 00:08:56,000 --> 00:08:59,000 or let me just go ahead and write embeddings. 169 00:08:59,000 --> 00:09:03,000 Dot I will go ahead and write embed underscore documents. 170 00:09:04,000 --> 00:09:08,000 So this is one of the functionality that I'm actually going to use. 171 00:09:08,000 --> 00:09:09,000 Embed underscore documents. 172 00:09:09,000 --> 00:09:16,000 If you want to see the definition it says that hey list of text words to embed inside this I have to 173 00:09:16,000 --> 00:09:19,000 give the list of texts that I really need to do the ending. 174 00:09:19,000 --> 00:09:22,000 So here I will just go ahead and create my list. 175 00:09:22,000 --> 00:09:22,000 Okay. 176 00:09:22,000 --> 00:09:30,000 Now with respect to the list list, let's say that I have two two sentences which I will be copying 177 00:09:30,000 --> 00:09:31,000 it over here. 178 00:09:31,000 --> 00:09:37,000 I'll say, hey, Alpha is the first character of the Greek alphabet and I'll say, hey, beta is the 179 00:09:37,000 --> 00:09:40,000 second character letter of the Greek alphabet. 180 00:09:40,000 --> 00:09:44,000 Okay, now what I will do is that I will just go ahead and execute this. 181 00:09:45,000 --> 00:09:45,000 Okay. 182 00:09:46,000 --> 00:09:49,000 So I'm just trying to embed this specific documents. 183 00:09:49,000 --> 00:09:52,000 And here I'm just going to see our one. 184 00:09:52,000 --> 00:09:54,000 And this is what I'm actually able to get it right. 185 00:09:54,000 --> 00:09:57,000 If I say hey our one of zero. 186 00:09:57,000 --> 00:09:59,000 So here this is my first one. 187 00:09:59,000 --> 00:10:02,000 So first over here you can see. 188 00:10:02,000 --> 00:10:07,000 And if you go ahead and see the length of this that many number of dimensions you will be able to see. 189 00:10:07,000 --> 00:10:14,000 So in short, over here, this gamma model actually creates a dimension of 2048 with respect to the 190 00:10:14,000 --> 00:10:15,000 vectors. 191 00:10:15,000 --> 00:10:15,000 Okay. 192 00:10:16,000 --> 00:10:19,000 Now let me go ahead and try one more functionality. 193 00:10:19,000 --> 00:10:20,000 So here I'm going to use the same embedding. 194 00:10:20,000 --> 00:10:22,000 And I'm going to embed a query. 195 00:10:22,000 --> 00:10:24,000 See initially I used embed documents. 196 00:10:24,000 --> 00:10:29,000 Right inside this I can give a list of text uh that I have here. 197 00:10:29,000 --> 00:10:31,000 I will be just giving one text. 198 00:10:31,000 --> 00:10:31,000 Right. 199 00:10:31,000 --> 00:10:32,000 What is one sentence? 200 00:10:32,000 --> 00:10:35,000 So what is the second letter of Greek alphabet? 201 00:10:35,000 --> 00:10:39,000 And here also you could see that second had asked beta is the second letter of Greek alphabet. 202 00:10:39,000 --> 00:10:45,000 And when I probably go ahead and see ah one of one, I'm actually able to get this particular, uh, 203 00:10:45,000 --> 00:10:46,000 vectors right. 204 00:10:47,000 --> 00:10:49,000 Uh, now this question is also some something similar only. 205 00:10:49,000 --> 00:10:50,000 Okay. 206 00:10:50,000 --> 00:10:53,000 Then I think, uh, both these vectors should match a bit. 207 00:10:53,000 --> 00:10:53,000 Right? 208 00:10:53,000 --> 00:10:56,000 Because here I'm just asking what is the second letter of Greek alphabet there? 209 00:10:56,000 --> 00:10:57,000 I've given the answer. 210 00:10:57,000 --> 00:11:01,000 So here if you go ahead and see just go ahead and compare this two vectors. 211 00:11:01,000 --> 00:11:05,000 So this is 2.18 here also you can see 2.35. 212 00:11:05,000 --> 00:11:08,000 So it's almost matching a bit right. 213 00:11:08,000 --> 00:11:12,000 That basically means you are talking more about these two sentences are almost similar because that 214 00:11:12,000 --> 00:11:17,000 is just an answer saying that hey beta is the second letter of the Greek alphabet. 215 00:11:17,000 --> 00:11:21,000 Here I've just asked the question as what is the second letter of the Greek alphabet? 216 00:11:21,000 --> 00:11:22,000 Right? 217 00:11:22,000 --> 00:11:26,000 So I hope you are able to understand about this whole Arma embeddings. 218 00:11:26,000 --> 00:11:32,000 Uh, similarly, you can go ahead and try different different embedding models also. 219 00:11:32,000 --> 00:11:32,000 Right. 220 00:11:32,000 --> 00:11:35,000 So Omar has different different embedding models okay. 221 00:11:35,000 --> 00:11:41,000 So for that in order to show you what all embedding models it has, I'm just going to open a new browser 222 00:11:41,000 --> 00:11:41,000 over here. 223 00:11:41,000 --> 00:11:43,000 And I'm going to probably hit this particular URL. 224 00:11:43,000 --> 00:11:46,000 So here it shows a blog embedding models. 225 00:11:46,000 --> 00:11:47,000 Right. 226 00:11:47,000 --> 00:11:49,000 So um, what are embedding models? 227 00:11:49,000 --> 00:11:53,000 Embedding models are models that are trained specifically to generate vector embeddings. 228 00:11:53,000 --> 00:11:54,000 Here you can see all the information. 229 00:11:54,000 --> 00:11:57,000 Now with respect to embedding models. 230 00:11:57,000 --> 00:11:59,000 You also have different different embedding models. 231 00:11:59,000 --> 00:12:00,000 Right. 232 00:12:00,000 --> 00:12:04,000 Let's say you have this mix by and embed large right. 233 00:12:04,000 --> 00:12:07,000 You want to go ahead and see this particular model. 234 00:12:07,000 --> 00:12:08,000 You can go ahead and see it over here. 235 00:12:08,000 --> 00:12:09,000 Right. 236 00:12:09,000 --> 00:12:12,000 Like how many parameters are there if you want to pull this okay. 237 00:12:12,000 --> 00:12:13,000 Let's go ahead and pull it. 238 00:12:13,000 --> 00:12:14,000 Let's see okay. 239 00:12:14,000 --> 00:12:17,000 So I'll paste it over here I'll press enter. 240 00:12:17,000 --> 00:12:20,000 Now you see if I'm downloading it downloading it for the first time. 241 00:12:20,000 --> 00:12:22,000 It may take some amount of time because this is 669 MB. 242 00:12:23,000 --> 00:12:24,000 So this is getting downloaded. 243 00:12:24,000 --> 00:12:27,000 Once it gets downloaded I can use this model. 244 00:12:27,000 --> 00:12:27,000 Okay. 245 00:12:27,000 --> 00:12:32,000 So uh, we will wait for some time since this is basically getting downloaded till then, we'll go ahead 246 00:12:32,000 --> 00:12:34,000 and explore other models also. 247 00:12:34,000 --> 00:12:37,000 So inside this you also have Nomic embed text. 248 00:12:37,000 --> 00:12:39,000 You also have all mini LM. 249 00:12:39,000 --> 00:12:42,000 So these are some very free models that are available. 250 00:12:42,000 --> 00:12:42,000 Right. 251 00:12:42,000 --> 00:12:47,000 And uh we can actually use this kind of models and perform our embeddings, uh, whatever things we 252 00:12:47,000 --> 00:12:48,000 really want to do. 253 00:12:48,000 --> 00:12:49,000 Okay. 254 00:12:49,000 --> 00:12:52,000 So for this also I will try to show you how to actually do it. 255 00:12:52,000 --> 00:12:58,000 And, uh, you know, uh, over here also you can see in alarm also you have this chroma DB if you want 256 00:12:58,000 --> 00:13:02,000 to save it in the form of collections, you can actually do that and again retrieve it and generate 257 00:13:02,000 --> 00:13:02,000 it. 258 00:13:02,000 --> 00:13:03,000 It is up to you. 259 00:13:03,000 --> 00:13:05,000 So we'll discuss more about this as we go ahead. 260 00:13:05,000 --> 00:13:09,000 But in this video as I said uh, that we are just focusing on the embedding techniques. 261 00:13:09,000 --> 00:13:12,000 Now let me just go back to my command prompt right now. 262 00:13:12,000 --> 00:13:17,000 228 to 2 2037 MB has been loaded. 263 00:13:17,000 --> 00:13:19,000 Now I will just go ahead and write my code. 264 00:13:19,000 --> 00:13:20,000 Till then. 265 00:13:20,000 --> 00:13:24,000 Okay, so now let's say that I want to use some other embedding models. 266 00:13:25,000 --> 00:13:28,000 Other embedding models okay. 267 00:13:28,000 --> 00:13:35,000 If you want to probably go ahead and explore the other embedding models, I'll keep the link over here 268 00:13:35,000 --> 00:13:36,000 so that you can refer it. 269 00:13:36,000 --> 00:13:37,000 Okay. 270 00:13:37,000 --> 00:13:39,000 So this is the link uh. 271 00:13:39,000 --> 00:13:41,000 Ah that is there for the other embedding models. 272 00:13:41,000 --> 00:13:48,000 Now once this basically gets downloaded the embedding model, I will just go ahead and call this okay. 273 00:13:48,000 --> 00:13:51,000 So here you can see I've created an embedding with all of my embedding. 274 00:13:51,000 --> 00:13:53,000 And I'm using the model MKB I embed large. 275 00:13:53,000 --> 00:13:55,000 Then I've said that this is the text. 276 00:13:55,000 --> 00:14:00,000 And we try to just find out what will be the my vector right. 277 00:14:00,000 --> 00:14:02,000 So this will basically be my query underscore result. 278 00:14:02,000 --> 00:14:06,000 Now let's quickly see how much time is basically involved. 279 00:14:06,000 --> 00:14:07,000 Another 200 MB. 280 00:14:07,000 --> 00:14:12,000 And then once this download actually happens it's it is hardly taking 38 second. 281 00:14:12,000 --> 00:14:13,000 So this is what right. 282 00:14:13,000 --> 00:14:19,000 If that model is not downloaded in your local then you have to first of all go ahead and download it. 283 00:14:19,000 --> 00:14:23,000 And here also you can see that another 120 MB is there. 284 00:14:23,000 --> 00:14:26,000 Uh, similarly that is with respect to all the LM models. 285 00:14:26,000 --> 00:14:31,000 If you do not have llama three, then again you have to go ahead and write llama pull llama three, 286 00:14:31,000 --> 00:14:33,000 otherwise your code will not get executed. 287 00:14:33,000 --> 00:14:34,000 Okay, but this. 288 00:14:34,000 --> 00:14:39,000 In short, what is happening is that this model is basically getting downloaded in the local itself. 289 00:14:39,000 --> 00:14:44,000 So another 3 to 4 seconds and then we are good to go to run this okay. 290 00:14:45,000 --> 00:14:48,000 So now here you can see it has downloaded it 11 KB. 291 00:14:48,000 --> 00:14:49,000 This is also downloaded. 292 00:14:49,000 --> 00:14:52,000 So it's just like downloading a Docker's right. 293 00:14:52,000 --> 00:14:54,000 Dockers also gets downloaded in a similar way. 294 00:14:54,000 --> 00:14:55,000 Right. 295 00:14:55,000 --> 00:14:57,000 So finally has got downloaded. 296 00:14:57,000 --> 00:14:58,000 Let's go ahead and execute it. 297 00:14:58,000 --> 00:15:05,000 Now I will be using this hole on my embedding with respect to MB, I embed large and here is your entire 298 00:15:05,000 --> 00:15:06,000 length. 299 00:15:06,000 --> 00:15:10,000 Now let's go ahead and see length of query result. 300 00:15:10,000 --> 00:15:14,000 So it has nothing but 1024 dimensions right? 301 00:15:14,000 --> 00:15:20,000 So I hope you are able to understand about the embedding technique using ulama. 302 00:15:20,000 --> 00:15:24,000 Uh, you can again try different, different open source embedding techniques. 303 00:15:24,000 --> 00:15:25,000 It is up to you. 304 00:15:25,000 --> 00:15:30,000 But again, our main aim was to understand how we can use ulama and perform embedding. 305 00:15:30,000 --> 00:15:35,000 But trust me, if you do not have APIs opening API, you can definitely use this. 306 00:15:35,000 --> 00:15:37,000 And uh, up going ahead, right? 307 00:15:37,000 --> 00:15:40,000 I'll also be creating an end to end project with the help of ulama. 308 00:15:40,000 --> 00:15:42,000 So yes, this was it. 309 00:15:42,000 --> 00:15:43,000 I will see you all in the next video. 310 00:15:43,000 --> 00:15:43,000 Thank you.