Soham Shah

3 followers · 3 following · 916 views

on the atlas — 49

highlights — 20

a QuestionAnsweredExtractor which generates question/answer pairs from a piece of text, and also a SummaryExtractor which extracts summaries, not only within the current text, but also within adjacent texts. We show that this allows for “chunk dreaming” - each individual chunk can have more “holistic” details, leading to higher answer quality given retrieved results.
Automated Metadata Extraction for Better Retrieval + Synthesis - LlamaIndex 🦙 0.8.29.post1
we first use the LLM to infer a set of metadata filters as well as the right query string to pass to the vector db (either can also be blank). This overall query bundle is then executed against the vector db. This allows for more dynamic, expressive forms of retrieval beyond top-k semantic search. The relevant context for a given query may only require filtering on a metadata tag, or require a joint combination of filtering + semantic search within the filtered set, or just raw semantic search.
Auto-Retrieval from a Vector Database - LlamaIndex 🦙 0.8.15
It’s tempting to think of explanations as a layer of polish on top of ideas. We believe that the best explanations are often something much deeper: they are interfaces to ideas, a way of thinking and interacting with a concept.
Distill Update 2018
stop occasionally and ask yourself: Am I working on what I most want to work on?" When you're young it's ok if the answer is sometimes no, but this gets increasingly dangerous as you get older.
How to Do Great Work
Just as we overestimate what we can do in a day and underestimate what we can do over several years, we overestimate the damage done by procrastinating for a day and underestimate the damage done by procrastinating for several years.
How to Do Great Work
it's possible to work too hard, and if you do that you'll find you get diminishing returns: fatigue will make you stupid, and eventually even damage your health. The point at which work yields diminishing returns depends on the type. Some of the hardest types you might only be able to do for four or five hours a day. Ideally those hours will be contiguous. To the extent you can, try to arrange your life so you have big blocks of time to work in. You'll shy away from hard tasks if you know you might be interrupted.
How to Do Great Work
The trouble with planning is that it only works for achievements you can describe in advance. You can win a gold medal or get rich by deciding to as a child and then tenaciously pursuing that goal, but you can't discover natural selection that way. I think for most people who want to do great work, the right strategy is not to plan too much. At each stage do whatever seems most interesting and gives you the best options for the future. I call this approach "staying upwind."
How to Do Great Work
When in doubt, optimize for interestingness. Fields change as you learn more about them. What mathematicians do, for example, is very different from what you do in high school math classes. So you need to give different types of work a chance to show you what they're like. But a field should become increasingly interesting as you learn more about it. If it doesn't, it's probably not for you.
How to Do Great Work
What should you do if you're young and ambitious but don't know what to work on? What you should not do is drift along passively, assuming the problem will solve itself. You need to take action. But there is no systematic procedure you can follow. When you read biographies of people who've done great work, it's remarkable how much luck is involved. They discover what to work on as a result of a chance meeting, or by reading a book they happen to pick up. So you need to make yourself a big target for luck, and the way to do that is to be curious. Try lots of things, meet lots of people, read lo…
How to Do Great Work
In general, gpt-3.5-turbo-0301 does not pay strong attention to the system message, and therefore important instructions are often better placed in a user message. If the model isn’t generating the output you want, feel free to iterate and experiment with potential improvements. You can try approaches like: Make your instruction more explicit Specify the format you want the answer in Ask the model to think step by step or debate pros and cons before settling on an answer
Chat completion - OpenAI API
laser focus on the next step in front of them combined with long-term vision. Most people only have one or the other.
Researchers and Founders
The best people in both groups spend a lot of time reflecting on some version of the Hamming question—"what are the most important problems in your field, and why aren’t you working on them?” In general, no one reflects on this question enough, but the best people do it the most, and have the best ‘problem taste’, which is some combination of learning to think independently, reason about the future, and identify attack vectors.
Researchers and Founders
“Your mind is for having ideas, not holding them” David Allen, author of Getting Things Done
Notion Second Brain
If we could improve in any way, how would we do it? What’s the No. 1 problem with our organization? Why? What’s not fun about working here? Who is really kicking ass in the company? Who do you admire? If you were me, what changes would you make? What don’t you like about the product? What’s the biggest opportunity that we’re missing out on? What are we not doing that we should be doing? Are you happy working here?
One On One | Andreessen Horowitz
A good practice is to have the employee send you the agenda in advance. This will give her a chance to cancel the meeting if nothing is pressing. It also makes clear that it is her meeting and will take as much or as little time as she needs. During the meeting, since it’s the employee’s meeting, the manager should do 10% of the talking and 90% of the listening.
One On One | Andreessen Horowitz
Yes, But vs. Yes, And This warm-up shows the power of building others’ ideas versus shooting them down. Taken from one of the principles of improv comedy, in this activity, you pair people and have them do the following:
The Best Design Thinking Exercises for Each Phase of a Project + Voltage Control
By 6 months, your baby will be starting to eat solid foods and this is the time to establish good eating habits. The nutrition your baby receives will set them up for a life of good health. Your baby will be making more sounds and be interested in things they can hold in their hands. Babies will smile easily and enjoy being around other people who are happy. They enjoy games of peek-a-boo, looking into the mirror and where objects are hidden for them to find.
Helping your baby grow from 0 to 5 years | Pregnancy Birth and Baby
They like to make a noise, and simple drums and shakers will provide them with a low-cost toy. Filling and emptying containers and supervised play crawling around the house are fun things for a child this age.
Helping your baby grow from 0 to 5 years | Pregnancy Birth and Baby
It is safe to place your baby on their tummy while they’re awake and you are with them. This strengthens their neck and encourages them to use their arms and kick their legs, which assists their physical development. But they should always sleep on their backs.
Helping your baby grow from 0 to 5 years | Pregnancy Birth and Baby
To model the hierarchical structure of KG, we map the en- tity and relation embeddings of the KG into the Poincar ́ e ball along with user and item embeddings. To the best of our knowledge, ours is the first work to consider knowledge-enhanced recommendation in the hyperbolic space
Knowledge-Enhanced Top-K Recommendation in Poincar ́e Ball