WEBVTT 1 00:00:00.660 --> 00:00:02.850 So when using Ollama 2 00:00:02.850 --> 00:00:05.820 through the command line, as I showed it to you earlier, 3 00:00:05.820 --> 00:00:07.290 you can chat with it, 4 00:00:07.290 --> 00:00:11.370 and you will build a chat history here, of course. 5 00:00:11.370 --> 00:00:13.890 What you can also do is configure it 6 00:00:13.890 --> 00:00:16.230 with help of various slash commands, 7 00:00:16.230 --> 00:00:19.170 and you can always type /? 8 00:00:19.170 --> 00:00:21.783 to see a list of the available commands. 9 00:00:23.100 --> 00:00:26.250 For example, you can use /show 10 00:00:26.250 --> 00:00:30.000 to view some information about the currently running model. 11 00:00:30.000 --> 00:00:32.010 With /show info, for example, 12 00:00:32.010 --> 00:00:36.000 you get a brief overview of the model architecture, 13 00:00:36.000 --> 00:00:39.450 so the model family, you could say, gemma3, in my case, 14 00:00:39.450 --> 00:00:41.130 the number of parameters, 15 00:00:41.130 --> 00:00:45.210 the available context length for this model, 16 00:00:45.210 --> 00:00:46.920 the quantization used, 17 00:00:46.920 --> 00:00:50.670 and some other pieces of information like, for example, 18 00:00:50.670 --> 00:00:53.130 the parameters that have been set 19 00:00:53.130 --> 00:00:56.640 for this running instance of the model, so to say, 20 00:00:56.640 --> 00:00:58.980 so the parameters that are currently active 21 00:00:58.980 --> 00:01:00.573 for this chat session. 22 00:01:01.560 --> 00:01:03.810 One word about the context length, by the way, 23 00:01:03.810 --> 00:01:06.150 this can be misleading because that is 24 00:01:06.150 --> 00:01:09.960 the actual context length supported by this model. 25 00:01:09.960 --> 00:01:11.730 It is not, however, 26 00:01:11.730 --> 00:01:14.850 necessarily the context length that has been set 27 00:01:14.850 --> 00:01:18.210 for your currently running chat session session. 28 00:01:18.210 --> 00:01:22.500 And I'll get back to how to control that in just a second. 29 00:01:22.500 --> 00:01:24.060 But that is something to note. 30 00:01:24.060 --> 00:01:26.970 This is just the maximum context length, 31 00:01:26.970 --> 00:01:28.893 not the currently active one. 32 00:01:29.880 --> 00:01:33.060 You also see the capabilities like text completion, 33 00:01:33.060 --> 00:01:35.670 and in case of the Gemma models also vision 34 00:01:35.670 --> 00:01:38.040 so that it can analyze images. 35 00:01:38.040 --> 00:01:39.450 And there's, of course, more 36 00:01:39.450 --> 00:01:42.591 you can inspect about this model. 37 00:01:42.591 --> 00:01:45.480 /show license shows you, well, the license, 38 00:01:45.480 --> 00:01:47.880 though here it fails to grab that. 39 00:01:47.880 --> 00:01:50.940 It's always a good idea to check Hugging Face for that then 40 00:01:50.940 --> 00:01:53.730 to make sure that you are respecting the license. 41 00:01:53.730 --> 00:01:57.690 And /show parameters shows you the currently set parameters. 42 00:01:57.690 --> 00:02:00.810 Now you might wonder who set parameters 43 00:02:00.810 --> 00:02:02.190 because you didn't. 44 00:02:02.190 --> 00:02:05.070 You did not see me set any parameters here. 45 00:02:05.070 --> 00:02:06.600 And for that you can go back to 46 00:02:06.600 --> 00:02:10.350 that Ollama model detail page on their website. 47 00:02:10.350 --> 00:02:14.820 And there you'll find this params area here. 48 00:02:14.820 --> 00:02:15.870 And if you click on that, 49 00:02:15.870 --> 00:02:19.290 you'll actually see a set of parameters, 50 00:02:19.290 --> 00:02:21.120 which will be different for each model, 51 00:02:21.120 --> 00:02:24.270 or which can be different for each model, I should say. 52 00:02:24.270 --> 00:02:28.380 And these parameters are simply set automatically 53 00:02:28.380 --> 00:02:30.450 Whenever you run a model. 54 00:02:30.450 --> 00:02:31.650 You can overwrite them, 55 00:02:31.650 --> 00:02:35.250 but these are default parameters that are set 56 00:02:35.250 --> 00:02:39.060 whenever you start a model with Ollama. 57 00:02:39.060 --> 00:02:40.800 And that's just something to be aware of 58 00:02:40.800 --> 00:02:45.510 and, therefore, this is what you also see here. 59 00:02:47.220 --> 00:02:48.053 By the way, 60 00:02:48.053 --> 00:02:51.840 I'll get back to this weird stop thing here later. 61 00:02:51.840 --> 00:02:54.900 You already know top_p, temperature, and top_k 62 00:02:54.900 --> 00:02:56.670 from the LM Studio section. 63 00:02:56.670 --> 00:02:57.810 In case you don't, 64 00:02:57.810 --> 00:02:59.880 attached you'll find a link to that lecture 65 00:02:59.880 --> 00:03:02.163 where I explain those parameters. 66 00:03:03.300 --> 00:03:05.880 What's also interesting is /show system 67 00:03:05.880 --> 00:03:07.650 which shows you the system prompt 68 00:03:07.650 --> 00:03:09.900 that was specified for this model. 69 00:03:09.900 --> 00:03:14.160 And here you see that no system message has been specified. 70 00:03:14.160 --> 00:03:16.170 That is, of course, something you can change. 71 00:03:16.170 --> 00:03:17.790 But here, as you see for this model, 72 00:03:17.790 --> 00:03:19.980 no system message has been set. 73 00:03:19.980 --> 00:03:21.240 And later you will learn 74 00:03:21.240 --> 00:03:23.670 how you can set a default system message 75 00:03:23.670 --> 00:03:26.100 for an entire model that will then apply 76 00:03:26.100 --> 00:03:29.313 to all chat sessions started based on that model.