I've been using Obsidian recently and I can say it's had a big impact on my life. When I was younger I used to read a lot, but i hardly remember anything. Now, when I read something important, I'm sure to summarize it in Obsidian. I can remember it easier and even if I can't, I know where to go to find that information.
You can use llamaindex to load your vault and parse it to a local LLM (you can use openai if you’re comfortable with it). You end up getting your own personal search engine and it’s amazing. I really should finish my write up about it (also written in Obsidian)
Do you know how this compares to [PrivateGPT](https://github.com/imartinez/privateGPT). I am honestly at the point of choice paralysis with all these new tools
Ditto. There's so many llms out there now. Is there a website where these are aggregated and ranked? Man, I've been too busy to keep on top of these developments.
It's more or less the same exact idea! Use langchain, import llm and embedding model, and query against it. The repo you provided does the same exact thing but using llama cpp python as the backend. I opted to write my own custom llm class with using textgen as the api backend so I can use the gpu since its way faster. But with the new cuBLAS support on llama cpp, it's a game changer so you can use either now. I do find the llama cpp + cuBLAS about 25% slower compared to pure GPU which is really good for what it is.
I get how there's so many choices nowadays and it's overwhelming but 95% of the repo you'll see just uses langchain. For the backend, llama cpp is your best bet minus the constant updates that break the quantized models. If you look for TheBloke on huggingface/reddit, you'll find all the best models. Look for the "ggml" ones which means its supported by llama cpp. But like I mentioned before, llama cpp has been doing so many model breaking changes so using TheBloke's models is your best bet because s/he updates really frequently. I personally prefer wizard-vicuna 13B, the uncensored one is pretty damn amazing.
Here's some example output how fast it is running 13B on a 3090 with a Ryzen 9 5900X
In [5]: output = llm("Q: Name the planets in the solar system? A: ", max_tokens=32, stop=["Q:", "\n"], echo=True)
llama_print_timings: load time = 209.23 ms
llama_print_timings: sample time = 11.39 ms / 32 runs ( 0.36 ms per token)
llama_print_timings: prompt eval time = 209.16 ms / 15 tokens ( 13.94 ms per token)
llama_print_timings: eval time = 1806.98 ms / 31 runs ( 58.29 ms per token)
llama_print_timings: total time = 3033.91 ms
In [6]: print(output)
{'id': '', 'object': 'text_completion', 'created': 1684604167, 'model': './models/Wizard-Vicuna-13B-Uncensored.ggml.q5_1.bin', 'choices': [{'text': 'Q: Name the planets in the solar system? A: 1. Mercury, 2. Venus, 3. Earth, 4. Mars, 5. Jupiter, 6. Saturn', 'index': 0, 'logprobs': None, 'finish_reason': 'length'}], 'usage': {'prompt_tokens': 15, 'completion_tokens': 32, 'total_tokens': 47}}
Mainly because it's open source and has a ton of functionality. I can bring text docs, diagrams, spreadsheets, images, audio, and even video under one single pkm. Because it's an offline app, it doesnt have any of the drawbacks of the online proprietary solutions.
I love Readwise Reader, and subscribed within a few days of trying out Reader Beta.
Despite my best intentions, though, I've not started to turn my highlights into flashcards, or started a daily review habit.
To GP: AIUI the workflow in Reader is meant to be (i) add to library, (ii) read and highlight, (iii) review highlights and create flashcards, (iv) review flashcards.
I don’t see why it wouldn’t apply to both? I just finished crime & punishment and could see how this summary strategy would help better understand the book.
I couldn’t disagree more. For me it’s much more important to have a broad knowledge and gain retention and comprehension by repeated readings of books I enjoyed. Writing summaries might be good, but certainly not at the expense of not reading as many new books
Inspectional reading is one of the key limitations for me when reading books on the Kindle. I need to be able to quickly flip around from toc to glossary and between chapters. There’s still too much friction for that with the ebook format.
Well it really depends what your goals are, right? What kind of books are you reading, to what end? I personally struggle to think of any book I have engaged with which I got value out of, that could be meaningfully summarized in a couple of hours. The author was an expert on the subject and they only managed to summarize it down to a few hundred pages - if there was only a couple of pages worth of ideas in there presumably they would have just written a blogpost.
An autobiography is not a novel, because it lacks the fundamental aspect of being fictional.
There is such a thing as an autobiographical novel, but that is not nonfiction.
Because novels are, by definition, works of fiction.
The closest thing to a ‘nonfiction novel’ is probably the historical novel, which is a fictionalized historical account… but I’m not sure why such a thing would particularly be suited to this summarization methodology.
"Fictional narrative: Fictionality is most commonly cited as distinguishing novels from historiography. However this can be a problematic criterion.
...
Several novels, for example Ông cố vấn written by Hữu Mai, were designed to be and defined as a "non-fiction" novel which purposefully recorded historical facts in the form of a novel.":
It's up to you. For me, I may write the parts I presently find most useful in detail. And some high-level summaries for reference. Then I can search my notes to know what book I need.
I'm still not sure why someone would need to do this? What kind of information are you obtaining from books that you later find yourself wanting to search for?
These kind of productivity tips always confuse me when they lack context of what the person writing is trying to do.
Like: Anki flashcards and spaced repetition. I see a lot of people advocating for them. I have never understood why because, in my personal experience, retaining lots of disconnected facts has never really been something I've needed to solve for. I do sort of see how it could work for rapid language learning (though I still think immersion, reading and writing are better than randomized vocabulary memorization, if you can take the time to do it that way), and I think I understand it in the context of fields like medicine where there's just a lot of facts you need to acquire, but people will advocate it for everything, and I just don't get it.
When I read a book, the process of reading it adds to the sum of my knowledge; I absorb the ideas, combine them with my own, and come to a new understanding of a topic. At that point, the book has accomplished its goal, for me.
But I acknowledge that's just how I read. I'm not an academic who might be later on finding myself needing to recall where I read something so I can cite it (although I find in general I can remember where I read certain things)... is that the use case here? What's the goal with building up an externalized knowledgeable?
Gist of it being, like you say, to not only summarise but to process into ones own knowledge base. And yes, this book is targeted at academics with the purpose of setting them up for producing papers, books, or other artefacts from that knowledge base.
> retaining lots of disconnected facts has never really been something I've needed to solve for.
At in regards to Anki usage, the “disconnected facts” seem to be something that only students (high school/college) or people new to Anki do. These people are very much using Anki for a short term goal like passing a test or class. For that use case, “disconnected facts” works just fine.
The people who use Anki (long term) outside of school and outside of 2nd language learning usually have much more connected webs of cards. While the individual cards are usually atomic, they might have a dozen or more cards all attacking a concept from different angles. Also in this usecase it seems just the act of formulating the questions is the biggest part of the learning. The reviewing part is almost secondary.
If I've looked at a concept enough to write cards attacking it from different angles... why would I need to see those cards again? At that point I know that concept.
That is indeed what advocates of spaced repetition say.
Personally I find that knowledge that’s internalized and used does not decay. And devoting effort to refreshing the DRAM for knowledge I’m not using seems like it fails the YAGNI test. Let my brain’s cache eviction algorithm do its thing. If I learned something once I figure I can probably learn it again.
One of my use cases is to memorize function names that I use about once a month - not often enough to memorize them from usage only, but frequently enough that the gains from not having to Google their name every time are worth it.
Facts I find interesting and relevant at the time I was reading them. As an example, the book "Invisible Women" has a lot of great data in it. I don't wanna know the exact numbers, but concepts like:
- The normal human dummy in crash tests define "normal" to be like a man, so women tend to have more injuries in car crashes.
- Women are more at risk for cardiac failure because their heart attacks go undiagnosed. Women experience different symptoms of heart attack than men, and most doctors have been trained using research done on males. For example, men feel it in their chest and left arm whereas women experience pain in their stomach.
The process of writing it or summarizing helps me remember these things so I don't need to look them up as often. But going through my notes again I saw I forgot some things! I don't know when I'll use them when I'm writing them, but I have referenced them conversations before.
I find that most of the things I read can be pretty effectively summarized in a single thesis sentence and a few additional sentences to generally describe how that thesis was supported. If you're only reading things that can't even be summarized in an abstract, you must be reading a lot of unopinionated biographies?
If after reading an entire book you haven’t already internalized its main thesis, it’s probably not a very valuable thesis.
But reducing a book down to a single thesis minimizes the takeaway value of the book, surely?
Take a book like Gödel Escher Bach. Sure, that book has a thesis. But the value of having read that book is not captured in ‘strange loops are all it takes to create beauty, complexity and consciousness’ - all the different ideas that underpin that thesis are what makes it valuable. That book lives as a set of new connections and pathways between ideas in my brain. And I read it over 20 years ago.
The vast majority of literature has much of its value in the way it is told, not just in the story it is telling. You can summarize Hamlet in a sentence or two, but that's not the reason it is still being read and performed 400+ years later.
Summarizing is not the same as condensing. Just jotting down your view helps you. Such that you don't even have to worry about making your version for an audience.
Strong agree from me, though what I do is less involved than the article. I normally just take a few per-chapter summary notes in an org-mode outline, in a filename that starts with the year I read the book.
Recently I started creating my own topic index with pointers to which books I found certain insights in. I know I could generate such a thing automatically with tagging, but I enjoy manually curating my index.
Pretty much what I do as well. I got a Notion account with links to all sorts of things from useful timestamps in videos to chapters in books and docs. It’s been incredibly helpful in contextualising my information when I gotta go out and actually do something.
I’m at 47 books read this year so far and mostly disagree with the author of the article. When you find a book that’s worth summarizing, do it. But I get exposed to many more ideas (and find the ones worth thinking about more) simply by reading more.
Also, I just like reading. My goal isn’t to have some kind of personal knowledge base to prove how much I’ve comprehended to myself.
Not in my experience. The knowledge here and on Reddit is shallow and does not explore any topic in depth. Theres a reason why Hegel took 1000+ pages to develop his logic. You can never get exposed to the full idea
> My goal isn’t to have some kind of personal knowledge base to prove how much I’ve comprehended to myself.
The reason many people accumulate wealth is to spend it, not to stare at it insecurely. It goes without saying that knowing more things expands the number of things you can do.
I also do about 50/year. It's entirely for recreation. A lot of the books are non-fiction and educational but I'm not doing work. I'm enjoying myself and there's no outcome I'm hoping for besides that enjoyment in the moment.
Sure, ChatGPT can do this. Just like you could drive a car over a marathon route. That’s not the point though, the point of summarizing is to force yourself to understand. It’s the journey not the destination that’s important.
I have been meditating for a few years now and recently decided to start building some custom meditations into my routine. I had a meditation for when I get out of the shower that helps me reflect on the day ahead. One that I listen to before work to let me put down distractions and think about what I want to accomplish today. Then I have one for after work to help me relax and separate my work mind from my home mind. I created these with the help of ChatGPT, and an AI TTS, and audacity to put it all together. I could have voiced it myself, but I found that it distracts me from the meditation when I think about my own voice.
Here’s a really good trick: Read a book then find the nearest victim to excitedly tell all about what you’ve learned. The back and forth conversation is even better than writing it down.
If you really found the book useful, go write a summary after that conversation. It will be a much stronger summary and also way easier to write.
I built a tool for myself for the purpose of grokking ideas from books called Emdash [1]. Over the years I've collected reams of highlights from books and articles but until recently, rarely reviewed or absorbed them. The core of this app uses on-device ML to show related passages with similar ideas from other books you've read, and I find that going broad and exploring concepts from different angles really helps in comprehension.
I'm testing out a summarization/rephrase feature backed by LLMs that you can try in the demo. In HN fashion I'm trying to build this openly and gather feedback to see what works. I'd like to push this further in the active direction the article mentions with something like a Socratic dialogue mode where you're nudged to re-explain and examine ideas.
If anyone uses this thing/has feedback, let me know. Source is available too [2].
I tried 'related' with a couple of passages. For the first passage, the first result was a good semantic match, but the rest were a little too far off. For the second passage, the results were amazing.
Perhaps for the first you just didn't have any more snippets that were closer?
Are the related snippets taken from a selection of snippets you created, or from the full text of other books?
A nice workflow might be to select a passage I'm reading in a book, and then see related passages from other books. But that requires I have DRM-free ebooks, and that these have already been chunked and indexed.
Yes, the demo mode is a random subset of things I've highlighted and it's heavily weighted around certain topics and sparse on others, so that's why some passages don't have the strongest semantic matches.
You're right that it would be nice to see things in situ as you're reading, but it would seem that most e-reading experiences are locked down. I appreciate the feedback!
I've been looking for something like this to review books. Two suggestions: 1. allow to load a book from a url, so notes could be added to arbitrary books; 2. allow to select text and add notes to that selection.
This is such a neat tool. The presentation is very pleasant. Is the intention to have the snippets/notes be shareable in the future? I actually made a similar tool [1] (though your's is much more complete), which I use to quickly find passages and relevant text when I'm blogging. And, I was thinking it might be really useful to have highly rated notes on a snippet available so that you can get someone else's insight on a particular selection. I'll give this a more in depth look later when I want to write another blog post.
Thanks for sharing, I like the name and I'll try it out. Yes, I'd like to add opt-in sharing features in the future -- seeing others' notes can be very insightful as you said.
This is cool. I imagine instructors creating instances for their classes so the whole class can engage with each other's notes.
I wonder if one could couple it with OCR so that you could point a phone at a page and drop into an emdash experience on the text that you've got a physical copy of. Or, you know, point it at your kindle so that your notes aren't locked into their ecosystem.
I'm building a backend that would support that kind of thing in a peer to peer kind of way (indexes content by piecewise hash so that you can recognize content you or your peers have annotations for and reattach those annotations despite differences in pagination, etc). If I ever get it into a demo-worthy state, I may reach out to see if we can make them work together.
Tiago Forte popularized the idea of "Progressive Summarization", which starts with extensive highlighting, then rounds of progressive refinement, ie bold key phrases and concepts in a given passage, then later review those and summarize in your own words. Strong approach for retention and comprehension...
If you're trying to answer questions, the even more advanced tactic is to read more books, but only read the parts that touch on your question. That's why god invented the subject index. Reading every word in order is not for people who have specific questions to answer, but for pleasure, or for learning the basics.
I took a course at UChicago on how to read a book based on Mortimer Adler’s book of the same name, and the takeaway is that the first reading of a book should be a quick one (table of contents, skimming, dipping into pages). Only after a book is shown to have promise are we then to engage in deep reading which not only involves summarizing but also syntopical reading, which is to read other books around the same topic.
One thing that has really helped me is to scribble notes in the margins and to write a précis after every chapter, then a précis on the inside cover of the book to summarize all my chapter précis.
Finally all good books should be read twice or more. Good reading is rereading. This is a hard rule to follow because nobody has time to read the same book twice but the truth is you can’t understand a book deeply on first reading because you don’t have the lay of the land and the benefit of retrospection. A rereading helps you focus on details missed the first time around. To be honest though, very few books meet my bar of my willing to reread them.
This may be cultural but I just cannot bring myself write in a book even in pencil. post-its notes for me this is a great idea if I have the discipline
I had the same problem but my instructor told me: a book devoid of scribbles is a book that hasn’t been engaged with deeply. If you have to, buy two copies of a book — one for display and another to scribble in, knowing that you’ll truly own the scribbled copy because you won’t be able to sell it.
Scholars from the middle ages wrote on the edges of their scrolls and codexes. Some of these notes are very interesting and useful to scholars. In elementary and primary school most are taught not to write in books because they have to be used for decades, but there is a long tradition of taking notes in books. They are very inexpensive now compared to the past and when you write in the you leave a little extra something for posterity.
The era is not the same. Our scribing in the margins is significantly less valuable than that of medieval scholars, because our mediums for recording, storing, archiving, and sharing information are significantly less restricted.
The levels of waste produced in the modern era are unimaginably higher than then, as well, so our responsibilities are completely different.
A book you won't write in is a book that owns you, rather than you owning the book. A book is designed to convey information on the page, writing in your book only adds to and fulfills it's purpose.
Good point, though I will say that I've checked out a lot of library books that have had useful markings people have put in them, though I wouldn't personally mark something that isn't my own property that way.
Your insistence on this kind of 'consumptive' and 'territory-marking' ownership of books is kind of silly IMHO.
I mean, suppose you have a 100 books. You're not reading more than 2 or 3 of them at once, if that; and it's not like you're doing anything with the rest other than keeping them on a shelf.
Now, Once you either have a family, or have friends with similar interests, or with kinds of similar interests etc - those 98 (or 100) books become something you lend, or even give away. And when you get older, you will pass on your entire library, at some point, or sell it off, or bequeath it etc. At any of these points, you want the book to be free of your personal jots and notes, and as close to pristine condition as possible, for the other readers.
Write your notes in a notebook, or on your PC or laptop (or whatever). They'll also be searchable and easily editable.
> And when you get older, you will pass on your entire library, at some point, or sell it off, or bequeath it etc. At any of these points, you want the book to be free of your personal jots and notes
While I generally have a communitarian orientation, I predict that as I approach this point, other end-of-life matters will occupy my headspace to a greater degree than any psychic rumblings about the odd marginalia in my dusty collection.
As an aside (because for the mere mortals among us it doesn’t apply) occasionally some historical inferences are made based on the reader’s margin notations.
I prefer to write on ruled paper and tuck it into the front cover for the following reasons:
- ruled paper is easier to write on
- you won't run out of space (just add sheets)
- you can rewrite summaries as you revisit books decades later and retuck the new summary
- you can quickly scan the papers to archive a digital copy
- you can remove your notes to lend the book (or not)
- you can transfer your notes to a new edition trivially
- you can be more candid in what you write, knowing it's trivially discarded without giving up the book
- all the notes are in one place and you don't have to thumb through the book looking for where you wrote an idea that you half-remember
- you have space to draw diagrams if appropriate
- you can find something on the internet that importantly relates, print off a page or two, and adjunct your notes
- you can do extensive internet research and build a whole matching folder of contents on your laptop, but you can print a page of Title/URL's to adjunct your notes as a backup (and for anyone who borrows the book+notes)
It's not clear from these articles how soon damage becomes apparent; I haven't noticed any yet in books which had Post-Its. It may not matter if you don't intend your books to have archival value.
I've tried Book Darts for marking pages, and they can also be used like paper clips to hold notes. I write short notes in pencil, but try to avoid dark leads which make marks hard to erase.
Used to feel the same way. But I've developed an appreciation for secondhand marked books, thinking about who was the previous reader. It kind of destigmatized it for me.
> Finally all good books should be read twice or more. Good reading is rereading.
Fully agreed!
> This is a hard rule to follow because nobody has time to read the same book twice
Disagreed. I read fiction and I tend to re-read a lot, sometimes three or four times. I read for pleasure, so I have the time. Reading is "me" time, and I spend it however I want.
It's only hard to re-read if you have some silly goal like "I must read N books this month". I find such goals worthless.
>It's only hard to re-read if you have some silly goal like "I must read N books this month". I find such goals worthless.
I'm on board. I have a great deal of fun hunting books and sitting down with the good ones I find. I don't care how many I go through or how many pages. Reading good stuff is a matter of quality of life for me. "A life without books is unlivable" Erasmus
Why stop at two? Susan Rigetti suggests four times! [1] I've never tried that but I fully agree with her :) and just quietly dream about regaining my focus for reading so I could reread books 4 times.
Quote to save people from falling into the rabbit hole of her blog:
> Over the years and after much formal and informal study, I’ve learned a pretty foolproof method for studying philosophy: read everything four times. Here’s how it works:
> - First read: Read casually, as if you’re reading a novel or a newspaper or magazine article. Your goal here is simply to observe, not to engage (yet).
> - Second read: This time, read to understand. Take notes. Ask yourself, “what does the author really mean here?” Summarize things in your own words. Try to break down the arguments being presented into bullet points, identifying the premises and the conclusions. When you read a term you are unfamiliar with or want to understand better, google it or look it up in the SEP.
> - Third read: Read again, and this time engage with the text. Go back to your notes, where you identified the arguments being presented. Think of arguments in favor of what the author is saying and arguments against. Think of counterexamples.
> - Fourth read: Now read for one last time. Read casually, the way you did in the first read. Notice how your understanding of the text is now so much richer and deeper than it was on the first read.
> If you study this way, you’ll walk away with an incredibly solid understanding of philosophy and an intellectual foundation that will serve you well for the rest of your life.
I totally agree, and surprised it's a bit controversial here. I understand that reading for pleasure can be different than reading for comprehension
But I lean towards reading for comprehension. I have a wiki which includes all the books I've read. The good ones get their own wiki pages with the main things I learned, and other notes.
A key point is that I don't take notes while I read. I only do it LATER -- because if you can't remember what to write down for a few days, then you probably won't retain it, and it may not be worth retaining.
----
As a specific example, when thinking about this -- I wonder if anybody has read "The Signal vs. The Noise" by Nate Silver? I remember reading it because a friend had a copy.
Many years later, off the top of my head, I can't remember a single thing in that book. Question for other readers: can you remember a single thing from it?
I think maybe it's because I kinda knew most of the stuff in that book? I will refer to my notes.
On the other hand, I've been re-reading Antifragile by Taleb, which I first read in 2012 I believe, and it struck me how many things I absorbed unconsciously from that book, which I didn't ascribe to it.
For example I remember talking about "hormesis" during COVID, i.e. small errors and stress. And also I took up some light weightlifting because I thought it was a good complement to bike riding. i.e. having 2 different kinds of exercise
---
edit: Just went to back to my notes on Silver's book (finished January 2014). Surprisingly I was extremely positive on the book -- I said the writing was engaging, it's well-sourced, very good set of topics (weather prediction, earthquakes, terrorists, poker, basketball, financial markets), and novel insights
Though the funny thing is that I didn't say what the actual insights were. The only one was the importance of knowing when you don't know, which I also got from Taleb
So maybe I should go back to that book and see if I still think it's good, and if there were insights I didn't get elsewhere
It's funny perhaps that my perception of the book has been clouded by Silver's reputation. I think he has been criticized for "predicting the present" and of "horse race coverage", and I think that's true now. But back when I read the book, I probably had a much more positive impression of him!! Very interesting
I'm guessing a large part of the controversy is related to the fact that reading has many different aspects. Possibly the article is referring to the same kind of reading you're referring to - technical books on various real-world topics.
However, it's much harder to see the value in doing anything like the article is suggesting for literature. Is there really much to gain from writing a summary of Virginia Woolf's To The Lighthouse, or of a book of poetry? Is that even the right way to engage with the material?
Oh definitely, and I think part of the pushback is framing it as "writing summaries". That sounds like homework. :)
But I would just call it engaging with the ideas after you read. So talking with a friend about a book is just as good, and even better.
And writing a comment on a forum can be just as good, as it's an active thing too. I sometimes write comments and then link them from my wiki.
I've had a GoodReads account in the past, but somehow it seems a bit shallow and artificial ... almost a bit like homework. It's better if the topics naturally come up, but if they don't, I still find writing notes to myself to enhance the reading experience.
I mean you may spend 1 week or 8 weeks reading a book. I don't see why spending 5 to 30 minutes writing notes a few days after can be viewed as bad.
I totally agree with the author in that those 5 or 30 minutes is way better spent writing about what you read, rather than 5 or 30 minutes reading something new that you may or may not retain.
And again I'm focusing on reading non-fiction for retention, not literature. (For better or worse, I read WAY more non-fiction than literature)
I strongly disagree. I read for enjoyment, even when I'm reading a more educational work; I don't wish to ruin my enjoyment by making myself summarize every book I read.
I disagree with this. Reading for the sake of summarizing takes the joy away from reading itself. I don't know anybody that'll read just because they want to read more books.
If I wanted to read summaries I would just read from Coles notes or Sparknotes, but both are essentially me just skimming headlines, and not getting to the juicy bits of the materials.
I don't think that this author's point is that we read for the sake of summarizing, per se, but rather that the act of summarizing forces us to engage more thoughtfully with what we've just read. It's that engagement itself that enriches our understanding of the book. By all means enjoy what you're reading while you're reading it!
That's the problem here -- for many books you don't need to develop a deep understanding of the book. Even for books in programming language, I may pick up a book because I am interested in the design of the language or a certain pattern of writing code, not because I want to use the language in the production or master the language, in which case understanding and thinking about pieces is good enough.
If you had a different purpose for reading the book, presumably you'd summarize it with that purpose in mind, depth where you need it, omission where you don't.
If you don't care to write that summary down... well that's kind of a separate thing.
165 comments
[ 2.8 ms ] story [ 204 ms ] threadI get how there's so many choices nowadays and it's overwhelming but 95% of the repo you'll see just uses langchain. For the backend, llama cpp is your best bet minus the constant updates that break the quantized models. If you look for TheBloke on huggingface/reddit, you'll find all the best models. Look for the "ggml" ones which means its supported by llama cpp. But like I mentioned before, llama cpp has been doing so many model breaking changes so using TheBloke's models is your best bet because s/he updates really frequently. I personally prefer wizard-vicuna 13B, the uncensored one is pretty damn amazing.
In [5]: output = llm("Q: Name the planets in the solar system? A: ", max_tokens=32, stop=["Q:", "\n"], echo=True)
llama_print_timings: load time = 209.23 ms llama_print_timings: sample time = 11.39 ms / 32 runs ( 0.36 ms per token) llama_print_timings: prompt eval time = 209.16 ms / 15 tokens ( 13.94 ms per token) llama_print_timings: eval time = 1806.98 ms / 31 runs ( 58.29 ms per token) llama_print_timings: total time = 3033.91 ms
In [6]: print(output) {'id': '', 'object': 'text_completion', 'created': 1684604167, 'model': './models/Wizard-Vicuna-13B-Uncensored.ggml.q5_1.bin', 'choices': [{'text': 'Q: Name the planets in the solar system? A: 1. Mercury, 2. Venus, 3. Earth, 4. Mars, 5. Jupiter, 6. Saturn', 'index': 0, 'logprobs': None, 'finish_reason': 'length'}], 'usage': {'prompt_tokens': 15, 'completion_tokens': 32, 'total_tokens': 47}}
It's a shame, because Obsidian is kinda neat software (although I'm not the note taking type myself).
Despite my best intentions, though, I've not started to turn my highlights into flashcards, or started a daily review habit.
To GP: AIUI the workflow in Reader is meant to be (i) add to library, (ii) read and highlight, (iii) review highlights and create flashcards, (iv) review flashcards.
Great source of interesting reads, too. So some potential prosocial utility there, too.
There is such a thing as an autobiographical novel, but that is not nonfiction.
Because novels are, by definition, works of fiction.
The closest thing to a ‘nonfiction novel’ is probably the historical novel, which is a fictionalized historical account… but I’m not sure why such a thing would particularly be suited to this summarization methodology.
https://en.m.wikipedia.org/wiki/Novel#Defining_the_genre
Same with Camus' The Stranger, or Sartre's Nausea.
These kind of productivity tips always confuse me when they lack context of what the person writing is trying to do.
Like: Anki flashcards and spaced repetition. I see a lot of people advocating for them. I have never understood why because, in my personal experience, retaining lots of disconnected facts has never really been something I've needed to solve for. I do sort of see how it could work for rapid language learning (though I still think immersion, reading and writing are better than randomized vocabulary memorization, if you can take the time to do it that way), and I think I understand it in the context of fields like medicine where there's just a lot of facts you need to acquire, but people will advocate it for everything, and I just don't get it.
When I read a book, the process of reading it adds to the sum of my knowledge; I absorb the ideas, combine them with my own, and come to a new understanding of a topic. At that point, the book has accomplished its goal, for me.
But I acknowledge that's just how I read. I'm not an academic who might be later on finding myself needing to recall where I read something so I can cite it (although I find in general I can remember where I read certain things)... is that the use case here? What's the goal with building up an externalized knowledgeable?
Gist of it being, like you say, to not only summarise but to process into ones own knowledge base. And yes, this book is targeted at academics with the purpose of setting them up for producing papers, books, or other artefacts from that knowledge base.
At in regards to Anki usage, the “disconnected facts” seem to be something that only students (high school/college) or people new to Anki do. These people are very much using Anki for a short term goal like passing a test or class. For that use case, “disconnected facts” works just fine.
The people who use Anki (long term) outside of school and outside of 2nd language learning usually have much more connected webs of cards. While the individual cards are usually atomic, they might have a dozen or more cards all attacking a concept from different angles. Also in this usecase it seems just the act of formulating the questions is the biggest part of the learning. The reviewing part is almost secondary.
Personally I find that knowledge that’s internalized and used does not decay. And devoting effort to refreshing the DRAM for knowledge I’m not using seems like it fails the YAGNI test. Let my brain’s cache eviction algorithm do its thing. If I learned something once I figure I can probably learn it again.
Surely using it is a repetition of it, essentially?
I always figured spaced repetition was a tool or help with knowledge you didn't have opportunity to apply otherwise
- The normal human dummy in crash tests define "normal" to be like a man, so women tend to have more injuries in car crashes.
- Women are more at risk for cardiac failure because their heart attacks go undiagnosed. Women experience different symptoms of heart attack than men, and most doctors have been trained using research done on males. For example, men feel it in their chest and left arm whereas women experience pain in their stomach.
The process of writing it or summarizing helps me remember these things so I don't need to look them up as often. But going through my notes again I saw I forgot some things! I don't know when I'll use them when I'm writing them, but I have referenced them conversations before.
But reducing a book down to a single thesis minimizes the takeaway value of the book, surely?
Take a book like Gödel Escher Bach. Sure, that book has a thesis. But the value of having read that book is not captured in ‘strange loops are all it takes to create beauty, complexity and consciousness’ - all the different ideas that underpin that thesis are what makes it valuable. That book lives as a set of new connections and pathways between ideas in my brain. And I read it over 20 years ago.
Recently I started creating my own topic index with pointers to which books I found certain insights in. I know I could generate such a thing automatically with tagging, but I enjoy manually curating my index.
Also, I just like reading. My goal isn’t to have some kind of personal knowledge base to prove how much I’ve comprehended to myself.
The reason many people accumulate wealth is to spend it, not to stare at it insecurely. It goes without saying that knowing more things expands the number of things you can do.
Yes and the reason many people accumulate wealth is to stare at it insecurely. What’s your point?
But also, try quizzing yourself about what you’ve read the day before. I bet the retention is slim.
Does anyone have tips for successfully building reflection into one's life? In the mean time I'll reflect on it.
Here is a sample one that I use to get out of work mindset: https://www.dropbox.com/s/r65gi7tk4str2h7/After%20Work%20Med...
If you really found the book useful, go write a summary after that conversation. It will be a much stronger summary and also way easier to write.
I'm testing out a summarization/rephrase feature backed by LLMs that you can try in the demo. In HN fashion I'm trying to build this openly and gather feedback to see what works. I'd like to push this further in the active direction the article mentions with something like a Socratic dialogue mode where you're nudged to re-explain and examine ideas.
If anyone uses this thing/has feedback, let me know. Source is available too [2].
[1] https://emdash.ai
[2] https://github.com/dmotz/emdash
Perhaps for the first you just didn't have any more snippets that were closer?
Are the related snippets taken from a selection of snippets you created, or from the full text of other books?
A nice workflow might be to select a passage I'm reading in a book, and then see related passages from other books. But that requires I have DRM-free ebooks, and that these have already been chunked and indexed.
You're right that it would be nice to see things in situ as you're reading, but it would seem that most e-reading experiences are locked down. I appreciate the feedback!
i like it a lot!
[1]: https://ishmael.app
I wonder if one could couple it with OCR so that you could point a phone at a page and drop into an emdash experience on the text that you've got a physical copy of. Or, you know, point it at your kindle so that your notes aren't locked into their ecosystem.
I'm building a backend that would support that kind of thing in a peer to peer kind of way (indexes content by piecewise hash so that you can recognize content you or your peers have annotations for and reattach those annotations despite differences in pagination, etc). If I ever get it into a demo-worthy state, I may reach out to see if we can make them work together.
Your content addressable system sounds very interesting, let me know when you have a demo.
I took a course at UChicago on how to read a book based on Mortimer Adler’s book of the same name, and the takeaway is that the first reading of a book should be a quick one (table of contents, skimming, dipping into pages). Only after a book is shown to have promise are we then to engage in deep reading which not only involves summarizing but also syntopical reading, which is to read other books around the same topic.
One thing that has really helped me is to scribble notes in the margins and to write a précis after every chapter, then a précis on the inside cover of the book to summarize all my chapter précis.
Finally all good books should be read twice or more. Good reading is rereading. This is a hard rule to follow because nobody has time to read the same book twice but the truth is you can’t understand a book deeply on first reading because you don’t have the lay of the land and the benefit of retrospection. A rereading helps you focus on details missed the first time around. To be honest though, very few books meet my bar of my willing to reread them.
Post-its work. For me I use a mechanical pencil.
The levels of waste produced in the modern era are unimaginably higher than then, as well, so our responsibilities are completely different.
A book you won't write in is a book that owns you, rather than you owning the book. A book is designed to convey information on the page, writing in your book only adds to and fulfills it's purpose.
I mean, suppose you have a 100 books. You're not reading more than 2 or 3 of them at once, if that; and it's not like you're doing anything with the rest other than keeping them on a shelf.
Now, Once you either have a family, or have friends with similar interests, or with kinds of similar interests etc - those 98 (or 100) books become something you lend, or even give away. And when you get older, you will pass on your entire library, at some point, or sell it off, or bequeath it etc. At any of these points, you want the book to be free of your personal jots and notes, and as close to pristine condition as possible, for the other readers.
Write your notes in a notebook, or on your PC or laptop (or whatever). They'll also be searchable and easily editable.
While I generally have a communitarian orientation, I predict that as I approach this point, other end-of-life matters will occupy my headspace to a greater degree than any psychic rumblings about the odd marginalia in my dusty collection.
As an aside (because for the mere mortals among us it doesn’t apply) occasionally some historical inferences are made based on the reader’s margin notations.
- ruled paper is easier to write on
- you won't run out of space (just add sheets)
- you can rewrite summaries as you revisit books decades later and retuck the new summary
- you can quickly scan the papers to archive a digital copy
- you can remove your notes to lend the book (or not)
- you can transfer your notes to a new edition trivially
- you can be more candid in what you write, knowing it's trivially discarded without giving up the book
- all the notes are in one place and you don't have to thumb through the book looking for where you wrote an idea that you half-remember
- you have space to draw diagrams if appropriate
- you can find something on the internet that importantly relates, print off a page or two, and adjunct your notes
- you can do extensive internet research and build a whole matching folder of contents on your laptop, but you can print a page of Title/URL's to adjunct your notes as a backup (and for anyone who borrows the book+notes)
(edit: formatting)
https://siarchives.si.edu/blog/post-it-or-not-post-it
http://libraries.ucsd.edu/preservation/postits.html
It's not clear from these articles how soon damage becomes apparent; I haven't noticed any yet in books which had Post-Its. It may not matter if you don't intend your books to have archival value.
I've tried Book Darts for marking pages, and they can also be used like paper clips to hold notes. I write short notes in pencil, but try to avoid dark leads which make marks hard to erase.
https://web.archive.org/web/20110128174227/http://libraries....
I do keep a notebook where I write about random stuff and a lot of thinking about what I’m reading goes in there.
That's the same book mentioned in the footnotes of the article.
"How to Read a Book: The Classic Guide to Intelligent Reading Mortimer J. Adler, Charles van Doren"
https://www.goodreads.com/book/show/567610.How_to_Read_a_Boo...
Fully agreed!
> This is a hard rule to follow because nobody has time to read the same book twice
Disagreed. I read fiction and I tend to re-read a lot, sometimes three or four times. I read for pleasure, so I have the time. Reading is "me" time, and I spend it however I want.
It's only hard to re-read if you have some silly goal like "I must read N books this month". I find such goals worthless.
I'm on board. I have a great deal of fun hunting books and sitting down with the good ones I find. I don't care how many I go through or how many pages. Reading good stuff is a matter of quality of life for me. "A life without books is unlivable" Erasmus
Quote to save people from falling into the rabbit hole of her blog:
> Over the years and after much formal and informal study, I’ve learned a pretty foolproof method for studying philosophy: read everything four times. Here’s how it works:
> - First read: Read casually, as if you’re reading a novel or a newspaper or magazine article. Your goal here is simply to observe, not to engage (yet).
> - Second read: This time, read to understand. Take notes. Ask yourself, “what does the author really mean here?” Summarize things in your own words. Try to break down the arguments being presented into bullet points, identifying the premises and the conclusions. When you read a term you are unfamiliar with or want to understand better, google it or look it up in the SEP.
> - Third read: Read again, and this time engage with the text. Go back to your notes, where you identified the arguments being presented. Think of arguments in favor of what the author is saying and arguments against. Think of counterexamples.
> - Fourth read: Now read for one last time. Read casually, the way you did in the first read. Notice how your understanding of the text is now so much richer and deeper than it was on the first read.
> If you study this way, you’ll walk away with an incredibly solid understanding of philosophy and an intellectual foundation that will serve you well for the rest of your life.
[1] https://www.susanrigetti.com/philosophy
But I lean towards reading for comprehension. I have a wiki which includes all the books I've read. The good ones get their own wiki pages with the main things I learned, and other notes.
A key point is that I don't take notes while I read. I only do it LATER -- because if you can't remember what to write down for a few days, then you probably won't retain it, and it may not be worth retaining.
----
As a specific example, when thinking about this -- I wonder if anybody has read "The Signal vs. The Noise" by Nate Silver? I remember reading it because a friend had a copy.
Many years later, off the top of my head, I can't remember a single thing in that book. Question for other readers: can you remember a single thing from it?
I think maybe it's because I kinda knew most of the stuff in that book? I will refer to my notes.
On the other hand, I've been re-reading Antifragile by Taleb, which I first read in 2012 I believe, and it struck me how many things I absorbed unconsciously from that book, which I didn't ascribe to it.
For example I remember talking about "hormesis" during COVID, i.e. small errors and stress. And also I took up some light weightlifting because I thought it was a good complement to bike riding. i.e. having 2 different kinds of exercise
---
edit: Just went to back to my notes on Silver's book (finished January 2014). Surprisingly I was extremely positive on the book -- I said the writing was engaging, it's well-sourced, very good set of topics (weather prediction, earthquakes, terrorists, poker, basketball, financial markets), and novel insights
Though the funny thing is that I didn't say what the actual insights were. The only one was the importance of knowing when you don't know, which I also got from Taleb
So maybe I should go back to that book and see if I still think it's good, and if there were insights I didn't get elsewhere
It's funny perhaps that my perception of the book has been clouded by Silver's reputation. I think he has been criticized for "predicting the present" and of "horse race coverage", and I think that's true now. But back when I read the book, I probably had a much more positive impression of him!! Very interesting
However, it's much harder to see the value in doing anything like the article is suggesting for literature. Is there really much to gain from writing a summary of Virginia Woolf's To The Lighthouse, or of a book of poetry? Is that even the right way to engage with the material?
But I would just call it engaging with the ideas after you read. So talking with a friend about a book is just as good, and even better.
And writing a comment on a forum can be just as good, as it's an active thing too. I sometimes write comments and then link them from my wiki.
I've had a GoodReads account in the past, but somehow it seems a bit shallow and artificial ... almost a bit like homework. It's better if the topics naturally come up, but if they don't, I still find writing notes to myself to enhance the reading experience.
I mean you may spend 1 week or 8 weeks reading a book. I don't see why spending 5 to 30 minutes writing notes a few days after can be viewed as bad.
I totally agree with the author in that those 5 or 30 minutes is way better spent writing about what you read, rather than 5 or 30 minutes reading something new that you may or may not retain.
And again I'm focusing on reading non-fiction for retention, not literature. (For better or worse, I read WAY more non-fiction than literature)
If I wanted to read summaries I would just read from Coles notes or Sparknotes, but both are essentially me just skimming headlines, and not getting to the juicy bits of the materials.
If you don't care to write that summary down... well that's kind of a separate thing.