Michael Moser MoserMichael

cooking up stuff

In memory of my dear mother, Vera. You taught me to be positive and be ready to learn from anyone. And your cooking was the best...

Maybe a bit rare to see cooking advice in a github gist, nevertheless - let's go! (I prefer recipes that do not require a lot of work)

légumes rôtis - roasted veggies (mostly cauliflower)

wash two big cauliflower heads, remove the blackened spots
separate the cauliflowers into pieces (medium sized, not not too small pieces) and cut off the big stems.

course on visual programming with Scratch

The aim here is to come up with a methodology/course of teaching the stuff, so the intended audience are teachers/parents who want to teach this stuff.

Scratch

A general intro to Scratch

at the bottom of the page there is a dropdown for choosing your language (very important to some people)

link to course

by Isa Fulford - she also wrote OpenAI cookbook and Andrew Ng

Intro

Base LLM trained on masive and diverse corpus of training text
Instruction tuned LLM a base LLM that has further been trained on instructions on how to achieve goals (like web searching). trining technique is reinforcement

learning about Charlie Chaplin

Chaplin managed to create cinema as an form of art! So here I am taking some notes, as I study his work.

Nowadays we have many of his his movies on youtube (as the copyright is expiring, no matter what - when I was a kid any appearance of Chaplin on TV was an event!).

Movies are compressed representation (as compared to books). They use additional visual/audio clues to tell a story - and to create a different kind of experience. Also the dynamics of movie scripts are quite different as compared to storytelling in a book. Someone had to figure out what works and what does not, and that guy was Chaplin (who did not have a formal education in any of these areas...)

His early work are two-reel movies. In many of them he is fighting a big bully adversary.
This doesn't work will full length feature movies: starting with 'The Kid', and later in all his movies in 'United Artist' - the adversary is something bigger, like poverty in 'the Kid', the system in 'Modern Times', fa

Job hunting tips

/More Eitzes from me, 'Eitzes is billik' - advice comes cheap.../

This gist is my experience. In general: do not trust such advice. Somehow the market tends to change every few years, always be aware of that possibility and be on the look-out.

Kinds of companies to work for

You can work for a big established company or for a startup. With more advanced companies you have a less stressed environment / and a more steady job, kind of.

https://learn.deeplearning.ai/courses/retrieval-augmented-generation

by Zain Hassan

Module1

RAG (Retrieval Augmented Generation) gives the LLM direct access to customer supplied documents - specialized knowledge, that wer not art of the training data. Not everything is known during model training, the missing links are

private databases (stuff that people/enterprises do not make public)
hard to access information

notes on: Agentic workflows course by Andrew Ng

https://learn.deeplearning.ai/courses/agentic-ai/lesson/pu5xbv/welcome!

Module1

the whole thing is about decomposing a workflow into a process with separate stages. A stage can either ask a LLM to do something or call an external tool (DB access, google query, etc)

(example: writing an essay, write outline, research th topic, write the essay)

	#!/bin/bash

	GITHUB_USER_NAME=$1

	if [[ "x${GITHUB_USER_NAME}" == "x" ]]; then
	cat <<EOF
	Usage: $0 <github-user-name>"

	Check out all gists or given github user - under the current directory.

	" Set text width as 72.

	# Collection of various stuff & advice

	# -------------------------

	Link to google in AI mode, with the option to chat with a LLM.
	Probably uses Gemini of various versions, the version can probably vary - according to task.

	https://www.google.com/search?udm=50

	On ubuntu: a date server with netcat, so that you can see something in a web browser

	# For a single run; access with http://localhost:12005
	d=$(date);echo -e "HTTP/1.1 200 OK\nConnection: close\nContent-Length: ${#d}\n\n${d}" \| nc -l -p 12005

	# For mutltiple runs; access with http://localhost:12005
	while true; do d=$(date);echo -e "HTTP/1.1 200 OK\nConnection: close\nContent-Length: ${#d}\n\n${d}" \| nc -l -p 12005; done

	# serving a video file