<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Blogposts on Blog of Kasra Darvish</title>
    <link>https://kasraprime.com/blog/blogposts/</link>
    <description>Recent content in Blogposts on Blog of Kasra Darvish</description>
    <generator>Hugo -- gohugo.io</generator>
    <language>en</language>
    <lastBuildDate>Thu, 08 Feb 2024 00:00:00 +0000</lastBuildDate><atom:link href="https://kasraprime.com/blog/blogposts/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Large Language Models</title>
      <link>https://kasraprime.com/blog/blogposts/large-language-models/</link>
      <pubDate>Thu, 08 Feb 2024 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/large-language-models/</guid>
      <description>They are called LLM for short. These are giant NLP models that are usually trained for text generation.</description>
    </item>
    
    <item>
      <title>Paper Localizing Active Objects From Egocentric Vision</title>
      <link>https://kasraprime.com/blog/blogposts/paper-localizing-active-objects-from-egocentric-vision/</link>
      <pubDate>Thu, 08 Feb 2024 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/paper-localizing-active-objects-from-egocentric-vision/</guid>
      <description>This paper is published in ACL and you can find it using this link.
It&amp;rsquo;s super interesting to see how different research projects are taking advantage of LLMs (GPT) by prompting them to do something and use those results to solve another problem.
Summary This paper tackles the problem of grounding (localizing) objects in images using text instructions by focusing on pre- and post-conditions of target objects; they call this active object grounding.</description>
    </item>
    
    <item>
      <title>Semi Live Journal</title>
      <link>https://kasraprime.com/blog/blogposts/semi-live-journal/</link>
      <pubDate>Thu, 01 Feb 2024 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/semi-live-journal/</guid>
      <description>This is going to be a live page about all sorts of things that I find beautiful. I intend to update this post at least once a month. For more regular updates check out my telegram channel: Darvishak Let&amp;rsquo;s get down to business.
Yet another text-to-video model &amp;lt;2024-02-01 Thu&amp;gt;
LUMIERE is a new text-to-video model by Google Research. Lumiere uses a space-time diffusion model for video generation</description>
    </item>
    
    <item>
      <title>Research Workflow</title>
      <link>https://kasraprime.com/blog/blogposts/research-workflow/</link>
      <pubDate>Tue, 08 Aug 2023 16:12:23 -0400</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/research-workflow/</guid>
      <description>In order to be good at something, you need to have good tools. You need to hone your crafts. Research is no excemption either. The following is a talk I gave at IRAL Lab at UMBC</description>
    </item>
    
    <item>
      <title>Machine Learning Experiment Tracking With Wandb</title>
      <link>https://kasraprime.com/blog/blogposts/machine-learning-experiment-tracking-with-wandb/</link>
      <pubDate>Tue, 08 Aug 2023 16:08:48 -0400</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/machine-learning-experiment-tracking-with-wandb/</guid>
      <description></description>
    </item>
    
    <item>
      <title>Research Skills</title>
      <link>https://kasraprime.com/blog/blogposts/research-skills/</link>
      <pubDate>Tue, 08 Aug 2023 15:58:51 -0400</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/research-skills/</guid>
      <description>Two of most important reserach skills that are usually not thought are networking and using AI tools. I presented the follwoing talk for the IRAL lab at UMBC.</description>
    </item>
    
    <item>
      <title>Art</title>
      <link>https://kasraprime.com/blog/blogposts/art/</link>
      <pubDate>Tue, 08 Aug 2023 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/art/</guid>
      <description></description>
    </item>
    
    <item>
      <title>Artificial Intelligence</title>
      <link>https://kasraprime.com/blog/blogposts/artificial-intelligence/</link>
      <pubDate>Tue, 08 Aug 2023 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/artificial-intelligence/</guid>
      <description>A branch of computer science which tries to create Intelligence in machines.</description>
    </item>
    
    <item>
      <title>ChatGpt</title>
      <link>https://kasraprime.com/blog/blogposts/chatgpt/</link>
      <pubDate>Tue, 08 Aug 2023 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/chatgpt/</guid>
      <description>An Artificial Intelligence chatbot and Nlp model by openAI.</description>
    </item>
    
    <item>
      <title>Classification</title>
      <link>https://kasraprime.com/blog/blogposts/classification/</link>
      <pubDate>Tue, 08 Aug 2023 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/classification/</guid>
      <description>Why Classification As Gendler states (2011), &amp;ldquo;classifying objects into groups allows us to proceed effectively in an environment teeming with overwhelming detail.&amp;rdquo;
Random Baseline In order to compute the accuracy of a random baseline for a multi-class classfication you can simply: \[acc = P(y=0) \times P(\hat{y}=0) + P(class is 1) \times P(you guess 1) \] where P(y=1) means the probability that the class is actually 1, and \(P(\hat{y}=1)\) means tne probability that you guess 1.</description>
    </item>
    
    <item>
      <title>Generative Models</title>
      <link>https://kasraprime.com/blog/blogposts/generative-models/</link>
      <pubDate>Tue, 08 Aug 2023 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/generative-models/</guid>
      <description>Generative AI models are a broad area in Machine Learning that instead of Classification focus on creating data. Classification is about prediction which is really cool; Nostradamus ;) However, the ability to create is superior; We become Gods, new kind of gods that really exist! Also, if you can create, you can predict.
With that introduction, we can jump into the edge of research in this field. This field is growing really fast and has gained a lot of attention from the public and media recently, especially the famous ChatGpt model by OpenAI.</description>
    </item>
    
    <item>
      <title>God</title>
      <link>https://kasraprime.com/blog/blogposts/god/</link>
      <pubDate>Tue, 08 Aug 2023 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/god/</guid>
      <description></description>
    </item>
    
    <item>
      <title>Intelligence</title>
      <link>https://kasraprime.com/blog/blogposts/intelligence/</link>
      <pubDate>Tue, 08 Aug 2023 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/intelligence/</guid>
      <description>It&amp;rsquo;s hard to define intelligence. I will do it later!</description>
    </item>
    
    <item>
      <title>Machine Learning</title>
      <link>https://kasraprime.com/blog/blogposts/machine-learning/</link>
      <pubDate>Tue, 08 Aug 2023 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/machine-learning/</guid>
      <description>Teaching Whenever you want to explain machine learning to someone, use the example of Teaching. Machine learning is like taking an exam from students. You compare students&amp;rsquo; results (model&amp;rsquo;s prediction) with the answers (labels) you have.
Resources https://course19.fast.ai/videos/?lesson=1 https://wandb.ai/aarora/weekly%5C_digest/reports/This-Week-in-Research-Papers--Vmlldzo2MTA1MDg?galleryTag=digests Generative vs Discriminative models https://stats.stackexchange.com/questions/12421/generative-vs-discriminative
Generative models provide a model of how the data is actually generated (think of them as giving you a model of both \(P(X|Y)\) and \(P(Y)\)), and discriminative algorithms as simply providing classification splits.</description>
    </item>
    
    <item>
      <title>Nlp</title>
      <link>https://kasraprime.com/blog/blogposts/nlp/</link>
      <pubDate>Tue, 08 Aug 2023 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/nlp/</guid>
      <description>Q: What is a joint probability?
We have formal languages like programming languages, and natural languages like English. Natural languages are not formally specified, therefore we need statistical models to learn from examples, and approximate the natural lang.
the number of times a term occurs in a document is called its term frequency
Language modeling is the task of assigning a probability to sentences in a language. The notion of a language model is inherently probabilistic.</description>
    </item>
    
    <item>
      <title>Teaching</title>
      <link>https://kasraprime.com/blog/blogposts/teaching/</link>
      <pubDate>Tue, 08 Aug 2023 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/teaching/</guid>
      <description>Teaching is an Art to me.</description>
    </item>
    
    <item>
      <title>Text-to-Video AI Models: A New Hype in Multimodal Machine Learning</title>
      <link>https://kasraprime.com/blog/blogposts/text-to-video-ai-models/</link>
      <pubDate>Fri, 07 Oct 2022 15:13:35 -0400</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/text-to-video-ai-models/</guid>
      <description>There has been great hype in multimodal research these past few days because of the release of multiple text-to-video models while text-to-image models are still pretty hype and not perfect yet. Diffusion models such as GLIDE in general and stable diffusion in particular are the state-of-the-art models for text-to-image tasks as of now. However, multiple papers came out recently that tackle the task of generating videos from text prompts or still images.</description>
    </item>
    
    <item>
      <title>Regex</title>
      <link>https://kasraprime.com/blog/blogposts/regex/</link>
      <pubDate>Sun, 12 Dec 2021 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/regex/</guid>
      <description>Basics Use this website interactively to learn. Use this website to generate regular expressions
Exclude something Use [^something] to exclude some patterns.
Include something Similar to excluding but [something].
Include one or another [one|another]
Repeats * 0 or more of something + 1 or more of something Some Characters . Everything except new line. in or der to search for [](.org) you need to write \[\[.*\]\]. Let&amp;rsquo;s say you want to keep the content between the brackets.</description>
    </item>
    
    <item>
      <title>Journal: Announcements about my blogging activity</title>
      <link>https://kasraprime.com/blog/blogposts/journal-announcement-2021-11-03/</link>
      <pubDate>Wed, 03 Nov 2021 23:39:40 -0400</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/journal-announcement-2021-11-03/</guid>
      <description>Some Announcement I am back at writing again. Don&amp;rsquo;t get me wrong, I still write everyday, but for myself, and it&amp;rsquo;s been a long time that I have not written anything publicly. Well, that&amp;rsquo;s about to change since I have some ideas to write about frequently. The theme of my frequent writings are the followings:
Research Paper Summaries Journals and Diary Book Summaries and Recommendations Research Paper Summaries The papers are going to be mostly in Machine Learning and Artificial Intelligence, but I might write about other fields occasionally; these other areas could be Neuroscience, Cognitive Science, and more.</description>
    </item>
    
    <item>
      <title>Note Taking</title>
      <link>https://kasraprime.com/blog/blogposts/note-taking/</link>
      <pubDate>Tue, 10 Aug 2021 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/note-taking/</guid>
      <description>Hey You Perfectionist Don&amp;rsquo;t spend endless hours contemplating the tools. Follow this rule:
As long as you have access to your data and they are palin text, you will be fine. The worst is that you can&amp;rsquo;t convert your files easily to another plain text format, but even in that case you can search your files and find the data you are looking for. Don&amp;rsquo;t think about publishing and making it look good, because notes evolve over time and you don&amp;rsquo;t want the pdfs of them you created two years ago.</description>
    </item>
    
    <item>
      <title>Blogging</title>
      <link>https://kasraprime.com/blog/blogposts/blogging/</link>
      <pubDate>Wed, 03 Mar 2021 23:39:40 -0400</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/blogging/</guid>
      <description>Tools Jekyll Edit files in BuildBlog folder. Do the following in terminal
cd BuildBlog export PATH=$HOME/.gem/ruby/2.6.0/bin:$PATH JEKYLL_ENV=production bundle exec jekyll build Then copy the \_site contents to blog folder in your github repo.
You should do this every time you add a new post.
Hugo Create the website (Once) hugo new site blog_name cd blog_name Add a new post hugo new posts/my-first-post.md Publish The Blog hugo server -D # To see it live hugo -D # To build the website Output will be in .</description>
    </item>
    
    <item>
      <title>How to apply</title>
      <link>https://kasraprime.com/blog/blogposts/2018-11-21-how-to-apply/</link>
      <pubDate>Wed, 21 Nov 2018 23:26:55 -0400</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/2018-11-21-how-to-apply/</guid>
      <description>In this post, I am going to talk about the process of applying for universities in United States of America, but most of the parts are general and applicable to other countries as well. I am assuming that you are an undergrad or Master&amp;rsquo;s student and want to apply for Master&amp;rsquo;s or Ph.D. degree. Before starting to apply for universities, you need to have a reasonable resume, which means you have to learn different skills and have some research experiences in your field of study.</description>
    </item>
    
    <item>
      <title></title>
      <link>https://kasraprime.com/blog/blogposts/editors-war/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/editors-war/</guid>
      <description></description>
    </item>
    
    <item>
      <title></title>
      <link>https://kasraprime.com/blog/blogposts/obsidian/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/obsidian/</guid>
      <description></description>
    </item>
    
    <item>
      <title></title>
      <link>https://kasraprime.com/blog/blogposts/second-brain-zettlekasten/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/second-brain-zettlekasten/</guid>
      <description></description>
    </item>
    
    <item>
      <title></title>
      <link>https://kasraprime.com/blog/blogposts/vim/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/vim/</guid>
      <description></description>
    </item>
    
    <item>
      <title></title>
      <link>https://kasraprime.com/blog/blogposts/vscode/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/vscode/</guid>
      <description></description>
    </item>
    
    <item>
      <title>Channelb</title>
      <link>https://kasraprime.com/blog/blogposts/channelb/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/channelb/</guid>
      <description>Type: #podcast
Scientist and the Sea (Episode 60) status: #DONE The story of Tommy Gregory Thompson a scientist who turns to a thief. He is known for his leading role in the rediscovery of SS Central America on September 11, 1988.
Silk Road Keywords: Tor Browser, Dark Web, Drugs, Libertarian
The story of Ross Ulbrik who created a dark net webiste to sell drugs. He was a proponent of libertarian and freedom and that was why he started this website.</description>
    </item>
    
    <item>
      <title>Emacs</title>
      <link>https://kasraprime.com/blog/blogposts/emacs/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/emacs/</guid>
      <description>Pros org mode extensible Powerful text editing capabilities Org agenda is really good Future proof Cons Emacs is single-threaded so a buggy plugin can freeze your emacs process. Not very fast Learn To search and open a file in all subdirectories from the current directory use Find file from here command or spc+f+F.
TODO how to open files in system&amp;rsquo;s default app form emacs? General Packagaes wc-mode to give you number of words in the file.</description>
    </item>
    
    <item>
      <title>Org Mode</title>
      <link>https://kasraprime.com/blog/blogposts/org-mode/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/org-mode/</guid>
      <description>Links Insert a link: C-c C-l and then choose org: for example Edit a link: Same as inserting a link To link to a url use the following syntax: [[LINK][DESCRIPTION]] Link to other file formats: You can simply use the syntax above for urls, but instead of a url, provide the path to a file. You can alternatively use the command org-insert-link or keyborad shortcut spc m l l. The default keybinding is C-c C-l.</description>
    </item>
    
    <item>
      <title>Podcasts List</title>
      <link>https://kasraprime.com/blog/blogposts/podcasts/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      
      <guid>https://kasraprime.com/blog/blogposts/podcasts/</guid>
      <description>Podcasts Name Status Genre Author Start Date Rating Episodes channelB #WIP Story, Fiction Ali Bandari 2021-04-13 5/5 Bplus #WIP Book Summary Ali Bandari 5/5 </description>
    </item>
    
  </channel>
</rss>
