开启时会模糊预览图，关闭后正常显示
Leopold's having fun, so here's my AI Safety twink portfolio. Total 1-year return: +892%. Criteria: Product advances human civilisation + good team. 50% Oklo (+1700%) 45% Tesla (+87%) 5% Nvidia (+55%)
I like using resources to bet on those making good moves in civilisation-level exponentials. It's easy alpha bc no one rly does it, but if you grasp how to build civilisations+how good teams work, then other factors are rounding errors in the long run.
Zoox's current service area. I don't think it's open for registration yet
i often see zoox mapping out hayes valley and closer to downtown, but i think it will be half a year or so before we can take rides there.
I rode in a Zoox (self-driving Amazon car) today! Overall an amazing experience; another rare time where the present truly looks, feels, and sounds like the future Waymo set a *very* high standard for me to compare to but I feel like Zoox was able to achieve a similar level of polish throughout the experience Pros: • App is great and actually works (rare!) • Driving is very smooth, about the same as a Waymo (both of which are nicer and smoother than most humans) • The leg space and seat layout are great and unparalleled. There's drink holders, usb ports, wireless charging, and music control on touch screen • 10/10 SFX, voice, animations, and subtle details. They matter! Cons: • Service area is small (mostly East Mission + SOMA, see below) • I feel a little less safe in Zoox versus Waymo. Waymo still has a driver's seat, a steering wheel, and more importantly, a dashboard which shows me what the car sees and what it is thinking and planning. Zoox has none of these (the touch screens don't show me LIDAR or AI predictions), which does feel super cool, but will take me a bit to get used to • The vehicle draws attention to itself more so than Waymo due to its irregularity, which I dislike because I don't feel safe in NE Mission and SOMA (but this isn't really Zoox's fault, is it? I trust it to not crash me!) • No Spotify integration but I could choose songs from Zoox FM Misc: • My pickup time was 7min, which is on par with or ~2min above my average Waymo pickup time • My ride was free. I don't know what the price will settle on when there's larger availability and also don't know current availability or timelines • There were two semi-urgent stops on my ride from a cyclist and a motorcyclist who both came aggressively close (around 1ft) to me. One stop was a bit less smooth than Waymo's would have been, but the car still made smart choices • Rides are available 8AM-11PM, riders must be 8 or older, no pets, alcohol, or smoking Overall: • Really cool experience. Feels less like taking a taxi and more like being in a smooth amusement park ride that tries (and succeeds!) at convincing me I'm living in the year 2030 • For now I'll stick with Waymo due to the UX, music control, familiarity, and availability (I've spent 105 hours of my life in a Waymo thus far apparently), but once Zoox expands their service area, especially if I'm with 2-3 friends, I'm looking forward to trying it more often Disclosures: • I love Waymo and self-driving cars • I own exactly one share in Amazon valued at $234 • I wasn't asked to tweet or review but no one told me I couldn't so send tweet
Photo I took of a Zoox afterwards for those who haven't seen it
biotech filter in claude is the worst AI filter i've ever seen in terms of false positives
they are simply preparing for the day when rl suddenly works
📦 Submit: • PR to bounty_004 with a runnable entrypoint • Results table (all MipNeRF360 scenes) • Visuals and tech brief • Dependencies + GPLv3-compatible licenses C++ preferred (Python = 20% award reduction). Deadline: Oct 12, 2025, 11:59 PM PST (Oct 12, 2025, 11:59 AM PST). Good luck!
Sponsors of total prize $2,430: @Auki $1000 @fulligin $500 @janusch_patas $300 @YeheLiu $280 @kennethlynne $200 @fhahlbohm $100 @mazy1998 $50 github issue: https://t.co/pwPZYpM3NU Discord: https://t.co/vHUPD8hcdg
🔔 Bounty #004 - Automatic Per-Scene Hyperparameter Optimization Build a system for gplv3 LichtFeld Studio that automatically finds optimal hyperparams per scene during training. Must use MCMC densification (Num Gaussians tunable). Target: +0.15 dB PSNR over MipNeRF360 baseline. Prize pool: $2,430. Details and link in the thread below!
🧾 Core Rules (Brief): • Execute during training without manual per-scene tuning. • Adjust learning rates for position, scale, rotation, opacity, and Spherical Harmonics. • Modify densification thresholds/intervals, number of iterations, number of Gaussians, and other quality/convergence parameters. • Fork from the bounty_004 branch. • Ensure runs are reproducible.