mvkel

17 ms·

421.
by mvkel 2y ago
It would be strange if they went about it this way, as it gives the godot community some unprecedented control over apple's release schedule
422.
by mvkel 2y ago
OpenAI has been a product company ever since ChatGPT launched. Their value is firmly rooted in how they wrap ux around models.
423.
by mvkel 2y ago
I'm not sure what your point is. If you already have a successful co that doesn't need to raise, then yeah you probably don't need to practice seed stage pitching. Also: > If you have a profitable company that needs to sc
424.
by mvkel 2y ago
this is actually a helpful little utility for someone applying to yc, or trying to raise. when refining a pitch, it takes a lot of, well, pitching to discover what's wrong with your message, or where it's falling flat. the more yo
425.
by mvkel 2y ago
> Turns out their ML/AI method doesn't generalize so well. I'd argue the opposite. AI typically generalizes very well. What it can't do well is specifics. It can't do the same thing over and over and follow every
426.
by mvkel 2y ago
I really wish people would stop evaluating a model's coding capability with one-shots. The vast majority of coding energy is what comes next. Even today, sonnet-3.5 is still the best "existing code base" model. Which is grati
427.
by mvkel 2y ago
This sounds like a business book on cats. Useful, yes, but not something I'd read for its writing value.
428.
by mvkel 2y ago
To me it just gets repetitive. After the first one, my brain recognizes the pattern. If chapter 1 starts with a bang, then fills in the blanks, then chapter 2 is structured in the exact same way, it feels formulaic; not good writing
429.
by mvkel 2y ago
This works for blog posts, certainly. But it falls apart if you're doing anything even slightly long form, or have multiple points to make. It's also why LinkedIn posts all sound the same. "It seemed like any other Monday. Li
430.
by mvkel 2y ago
I'm sure they're still deploying years of loss carry-forwards. Nothing wrong with that, but I doubt they're generating tax revenue yet. That'll eventually change of course, and they'll be the best space game in town
431.
by mvkel 2y ago
The 4-star system is indeed wonderful. A rating that is truly "neutral" doesn't exist in a world of human subjectivity
432.
by mvkel 2y ago
I think this release is for the researchers who worked on it and would quit if it never saw daylight
433.
by mvkel 2y ago
That looks... out there. Is it parked at a Lagrange point?
434.
by mvkel 2y ago
Very strange. Almost like the altimeter setting wasn't set to the right number?
435.
by mvkel 2y ago
> next door to zero interest I now use this a few times a month, thanks to this anecdote
436.
by mvkel 2y ago
Sure, and there's a sprinter who can outrun Usain Bolt who just happened to become an accountant instead
437.
by mvkel 2y ago
> for purposes other than workout programming I specifically called out workouts. one tedious thing is needing to manually create the workouts. Unless you're doing the same thing week in and week out, it gets old quickly. I'd b
438.
by mvkel 2y ago
it always surprises me how much data is locked up in garmin's (ancient) sdk. for example, the FIT file format, used exclusively for programming workouts on all bike computers, are only supported in one or two apps! I suppose that'
439.
by mvkel 2y ago
The biggest delta over regular o1 that I've seen is asking it to make a PRD of an app that I define as a stream-of-consciousness with bullet points. It's fantastic at finding needles in the haystack, so the contradictions are none
440.
by mvkel 2y ago
This is probably the right way to think about it. O1-pro is an absolute monster when it comes to architecture. It is staggering the breadth and depth that it sees. Ask it to actually implement though, and it trips over its shoelaces almost
441.
by mvkel 2y ago
Going to try this! Thanks for the tip
442.
by mvkel 2y ago
If it's a context issue, it's an issue with how cursor itself sends the context to these reasoning LLMs. Context alone shouldn't be the reason that sonnet succeeds consistently, but others (some which have even bigger context
443.
by mvkel 2y ago
Thanks for all the work you do! v0 is magical. I absolutely love the feature where I can add a chunky component that v0 made to my repo with npx
444.
by mvkel 2y ago
My point was that a prompt that simple could be held and executed very well by sonnet, but all other models (especially reasoning models) crash and burn. It's a 15 line tsx file so context shouldn't be an issue. Makes me wonder if
445.
by mvkel 2y ago
o1-pro is incredibly good at this. You'll be amazed
446.
by mvkel 2y ago
I've been using cursor since it launched, sticking almost exclusively to claude-3.5-sonnet because it is incredibly consistent, and rarely loses the plot. As subsequent models have been released, most of which claim to be better at cod
447.
by mvkel 2y ago
Bring back .plan files!
448.
by mvkel 2y ago
I'd argue that you could write only for yourself for your entire life and still have a successful blog.
449.
by mvkel 2y ago
This is incredibly fascinating. I feel like one round of RL could potentially fix "short circuits" like these. It seems to be convinced that a particular rule isn't "allowed," when it's totally fine. Wouldn&#x2
450.
by mvkel 2y ago
A great book on failure is "Riding the Runaway Horse," about the rise and fall of Wang Computers. A cautionary tale about hubris and nepotism. https://www.amazon.com/Riding-Runaway-Horse-Decline-Laborato... It is

More ›