Can you go into a little more detail? What is broken exactly?
809 karma · joined January 5, 2019
Visual-Focused Algorithms Cheat Sheet — https://photonlines.substack.com/p/visual-focused-algorithms...
Visual Data Structures Cheat Sheet — https://photonlines.substack.com/p/visual-data-structures-ch...
Ilya Sutskever: Yeah, so the thing is the transformer is a combination of multiple ideas simultaneously of which attention is one.
Lex Friedman: Do you think attention is the key?
Ilya Sutskever: No, it's a key, but it's not the key. The transformer is successful because it is the simultaneous combination of multiple ideas. And if you were to remove either idea, it would be much less successful. So the transformer uses a lot of attention, but attention existed for a few years. So that can't be the main innovation. The transformer is designed in such a way that it runs really fast on the GPU. And that makes a huge amount of difference. This is one thing. The second thing is that transformer is not recurrent. And that is really important too, because it is more shallow and therefore much easier to optimize. So in other words, it uses attention, it is a really great fit to the GPU and it is not recurrent, so therefore less deep and easier to optimize. And the combination of those factors make it successful.
DeepSeek (with thinking & reasoning & search turned on)
Claude Code
QWEN Coder (with thinking & reasoning & search turned on)
ChatGPT
Google Gemini
- DeepSeek: excellent at coming up with solutions and churning out prototypes / working solutions with Reasoning mode turned on.
- Claude Code: I use this with cursor to quickly come up with overviews / READMEs for repos / new code I'm browsing and in making quick changes to the code-base (I only use it for simple tasks and don't usually trust it for implementing more advanced features).
- QWEN Coder: similar to deep-seek but much better at working with visual / image data sets.
- ChatGPT: usually use it for simple answers / finding bugs in code / resolving issues.
- Google Gemini: is catching up to other models when it comes to coding and more advanced tasks but still produces code that is a bit too verbose for my taste. Still solid progress since initial release and will most likely catch up to other models on most coding tasks soon. 1. "Work closely with T-Mobile"
2. "Prioritize touch UI development, simplifying basic functionality and PC suite development very high."
4. "Analyse what could be Apple’s next release of “iPhone mini” to mass market price points and plan counter-measures for it."
5. "Kill market for such an expensive device by filling mid-range with own/Google/Yahoo experiences"
6. "Accelerate Nokia's own free push e-mail project and make it less hidden within the company."
7. "Investigate and play hard in possible IPR infringements"
8. "Drive key partnerships to highlight Nokia's superior strength in the market, keeping things in perspective."
10. "Highlight potential weaknesses of the iPhone"
This whole report shows exactly why Nokia failed.
"When he ascends to his attic, she does not cling to his coat tails. When he wishes to be alone, she completely eliminates herself from his life. She spares him disharmonious contacts and protects the serenity of his mind with the devotion of a vestal virgin guarding the sacred fire.
It is by no means impossible that with a less-sacrificing mate, Einstein would not have made the discoveries which link his name with the immortals. Thus love, that moves the sun and all the stars, sustains in its lonely path the genius of Albert Einstein."
"A prospective study of dietary intake of vitamin D found women with daily intake above 400 IU had a 40% lower risk of MS. [53] In a study among healthy young adults in the US, White men and women with the highest vitamin D serum levels had a 62% lower risk of developing MS than those with the lowest vitamin D levels. [54]"
"Another prospective study in young adults from Sweden also found a 61% lower risk of MS with higher serum vitamin D levels; [55] and a prospective study among young Finnish women found that low serum vitamin D levels were associated with a 43% increased risk of MS. [56] In prospective studies of persons with MS, higher vitamin D levels have been associated with reduced disease activity and progression. [57,58]"
"Collectively, the current evidence suggests that low vitamin D may have a causal role in MS and if so, approximately 40% of cases may be prevented by correcting vitamin D insufficiency. [59] This conclusion has been strengthened substantially by recent evidence that genetically determined low levels of vitamin D predict higher risk of multiple sclerosis."
"While this may largely be due to genetic differences, some studies suggest that T1D rates are lower in sunnier areas. Early evidence suggesting that vitamin D may play a role in T1D comes from a 30-year study that followed more than 10,000 Finnish children from birth: Children who regularly received vitamin D supplements during infancy had a nearly 90% lower risk of developing type 1 diabetes than those who did not receive supplements. [61]"
"A prospective study among healthy young adults in the US found that White individuals with the highest levels of serum vitamin D had a 44% lower risk of developing T1D in adulthood than those with the lowest levels. [63]"
"The Vitamin D and Omega 3 trial (VITAL), a randomized double-blind placebo-controlled trial following more than 25,000 men and women ages 50 and older, found that taking vitamin D supplements (2,000 IU/day) for five years, or vitamin D supplements with marine omega-3 fatty acids (1,000 mg/day), reduced the incidence of autoimmune diseases by about 22%, compared with a placebo. Autoimmune conditions observed included rheumatoid arthritis, psoriasis, polymyalgia rheumatica, and autoimmune thyroid diseases (Hashimoto’s thyroiditis, Graves’ disease). [80]"
2. This once again depends on the company. I prefer not interfacing with them: my job is to deliver value to the business by serving both the staff and the businesses' customers. My number one initiative when I am on-boarded to any organization is to understand: what is your goal, what do you want to accomplish and who are you here to serve? Once I figure that out, I can think in terms of end-users and various stake-holders and try to work backwards rather than having to have many useless meetings and having to constantly wonder what I should do next. Think in terms of the customer: what can you do to improve the product for them and what can we do to improve the business in general?
3. I usually do yes, but most of them are led by upper management and I mostly think of them as a waste of time. If I need anything, I get in touch to the user or customer who has the information and find this information directly. There is one recurring meeting that I do tend to have with my team: weekly learning sessions and these are non-mandatory and not really recurring.
4. Documentation to me is very important, so any communication which I have to do more than twice I document and I put it on a wiki or shared knowledge base which both me and the team can reference and point to in case anyone needs guidance. I'm also a huge fan of written requirements but that's me.
5. 1) Minimize meetings 2) Unblock your team by helping them out when they need help and teaching them how to solve their own issues instead of always having to come to you - if they can't resolve problems themselves or they come to you every day, maybe it's time to think about letting them go and hiring someone who is able to take more ownership and has a bit more talent in solving problems 3) Keep asking the question why? What is the business trying to achieve with my team and how can I improve it? Which activities can I do which are 'high-leverage' and which will help us accomplish what we need to accomplish? I'm very hands on so the way that I play 'lead' is by minimizing micro-management and getting my hands dirty: I develop the hard stuff that I believe my team won't be able to handle and try to also foster communication between my team as well: I tell them if they ever need anything to come to me directly and I will try to get them the information they need. Communication is the number one thing that will determine a team's success so I try to keep it moving - how can we communicate what we need to develop more effectively and how do we make this process more efficient? Sprints and stories and all that other crap to me is a sign that most software firms today are broken and that most developers take 0 ownership on aligning with the business. This is the reason why we have so much broken and shitty software today as well. Hire people that are excited about developing solutions for real human beings and that WANT to see the business progress without having a 'product owner' or 'sprint planner' or some bullshit-job personnel having to either disturb you and guide you and teach your staff to do the same. Also, take ownership and pride in doing quality work and teach your team to do the same by leading by example.
6. Make sure to understand the product and business and first and foremost: the customer and do your best to solve their problems. Also, read books which aren't just about copying stories on a story board or about shitty broken agile processes; understand general things about business and human psychology and usability and design - this will equip you to make decisions without having to be micro-managed by 'the business.' If the business needs and insists on micro-managing you or your team - then leave the company and find a place that isn't broken (albeit it is tough to find these...when you do find them, you'll find a lot of rewards in working for them - autonomy and pride and not being micro-managed are all amazing things).